def main():
"""Semi Automated ML App with Streamlit """
activities = ["EDA", "Plots", "Model Building", "About"]
choice = st.sidebar.selectbox("Select Activities", activities)
if choice == 'EDA':
st.subheader("Exploratory Data Analysis")
data = st.file_uploader("Upload a Dataset", type=["csv", "txt"])
# file_buffer = st.file_uploader("Upload a Dataset", type=["csv", "txt"])
# data = io.TextIOWrapper(file_buffer)
if data is not None:
df = pd.read_csv(data)
st.dataframe(df.head())
if st.checkbox("Show Shape"):
st.write(df.shape)
if st.checkbox("Show Columns"):
all_columns = df.columns.to_list()
st.write(all_columns)
if st.checkbox("Summary"):
st.write(df.describe())
if st.checkbox("Show Selected Columns"):
selected_columns = st.multiselect("Select Columns",
all_columns)
new_df = df[selected_columns]
st.dataframe(new_df)
if st.checkbox("Show Value Counts"):
st.write(df.iloc[:, -1].value_counts())
if st.checkbox("Correlation Plot(Matplotlib)"):
plt.matshow(df.corr())
st.pyplot()
if st.checkbox("Correlation Plot(Seaborn)"):
st.write(sns.heatmap(df.corr(), annot=True))
st.pyplot()
if st.checkbox("Pie Plot"):
all_columns = df.columns.to_list()
column_to_plot = st.selectbox("Select 1 Column", all_columns)
pie_plot = df[column_to_plot].value_counts().plot.pie(
autopct="%1.1f%%")
st.write(pie_plot)
st.pyplot()
elif choice == 'Plots':
st.subheader("Data Visualization")
data = st.file_uploader("Upload a Dataset", type=["csv", "txt"])
if data is not None:
df = pd.read_csv(data)
st.dataframe(df.head())
if st.checkbox("Show Value Counts"):
st.write(df.iloc[:, -1].value_counts().plot(kind='bar'))
st.pyplot()
# Customizable Plot
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox(
"Select Type of Plot",
["area", "bar", "line", "hist", "box", "kde"])
selected_columns_names = st.multiselect("Select Columns To Plot",
all_columns_names)
if st.button("Generate Plot"):
st.success("Generating Customizable Plot of {} for {}".format(
type_of_plot, selected_columns_names))
# Plot By Streamlit
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
# Custom Plot
elif type_of_plot:
# fig, ax = plt.subplots()
# ax.scatter(df[selected_columns_names])
# st.pyplot(fig)
cust_plot = df[selected_columns_names].plot(
kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
elif choice == 'Model Building':
st.subheader("Building ML Models")
data = st.file_uploader("Upload a Dataset", type=["csv", "txt"])
if data is not None:
df = pd.read_csv(data)
st.dataframe(df.head())
# Model Building
X = df.iloc[:, 0:-1]
Y = df.iloc[:, -1]
seed = 7
# prepare models
models = []
models.append(('LR', LogisticRegression()))
models.append(('LDA', LinearDiscriminantAnalysis()))
models.append(('KNN', KNeighborsClassifier()))
models.append(('CART', DecisionTreeClassifier()))
models.append(('NB', GaussianNB()))
models.append(('SVM', SVC()))
# evaluate each model in turn
model_names = []
model_mean = []
model_std = []
all_models = []
# all_columns = []
scoring = 'accuracy'
for name, model in models:
kfold = model_selection.KFold(n_splits=10, random_state=seed)
cv_results = model_selection.cross_val_score(model,
X,
Y,
cv=kfold,
scoring=scoring)
model_names.append(name)
model_mean.append(cv_results.mean())
model_std.append(cv_results.std())
accuracy_results = {
"model name": name,
"model_accuracy": cv_results.mean(),
"standard deviation": cv_results.std()
}
all_models.append(accuracy_results)
if st.checkbox("Metrics As Table"):
st.dataframe(
pd.DataFrame(zip(model_names, model_mean, model_std),
columns=["Algo", "Mean of Accuracy", "Std"]))
if st.checkbox("Metrics As JSON"):
st.json(all_models)
if option=="sensor":
sensor_data=load_sensor(20)
sensor_data1=load_sensor(100000)
if st.sidebar.checkbox("Sensor dataset"):
st.subheader("Sensor datasets")
st.write(sensor_data1)
dataframe = pd.DataFrame({
'first column': ["sensor_00","sensor_01","sensor_02"]})
st.sidebar.subheader("Choose your sensor for ploting")
sensor_option = st.sidebar.selectbox('',dataframe['first column'])
# print(sensor_data[sensor_option])
# print(sensor_data.columns)
#df3=pd.DataFrame({"date":sensor_data["timestamp"],sensor_option:sensor_data[sensor_option]})
if st.sidebar.button("Plot"):
#df3=df3.rename(columns={'date':'index'}).set_index('index')
st.subheader("Line chart")
st.line_chart(sensor_data[sensor_option])
st.subheader("Bar chart")
st.bar_chart(sensor_data[sensor_option])
sensor_data[sensor_option].hist()
st.subheader("Histogram")
st.set_option('deprecation.showPyplotGlobalUse', False)
st.pyplot()
st.subheader("Area chart")
st.area_chart(sensor_data[sensor_option])
#st.line_chart(sensor_data[sensor_option])
#st.subheader("Area chart")
#st.area_chart(sensor_data[sensor_option])
"field": "a",
"type": "quantitative"
},
"y": {
"field": "b",
"type": "quantitative"
},
},
}
# 5 empty charts
st.vega_lite_chart(spec)
st.pyplot()
st.line_chart()
st.bar_chart()
st.area_chart()
# 1 empty map
st.deck_gl_chart()
# 6 errors
try:
st.vega_lite_chart({})
except Exception as e:
st.write(e)
try:
st.vega_lite_chart(data, {})
except Exception as e:
st.write(e)
def classify_data(dataset, data_type):
"""Classify data and health risks with selected classification."""
filename = "/home/maddykapfhammer/Documents/Allegheny/MozillaFellows/predictiveWellness/vigor/dataFiles/PubMedArticles.csv"
naive_interpretation = ""
gini_interpretation = ""
entropy_interpretation = ""
svm_interpretation = ""
amount = st.number_input("How much data would you like to be produced?", min_value=1)
st.header("Please Choose Your Method of Classification:")
naive_classification = st.checkbox("Naive Bayes Classification")
gini_classification = st.checkbox("Gini Index Decision Tree Classification")
entropy_classification = st.checkbox("Entropy Decision Tree Classification")
svm_classification = st.checkbox("Support Vector Machine Classification")
if naive_classification:
with st.spinner("Classifying with Naive Bayes..."):
new_data = naive_bayes.import_data(data_type)
st.area_chart(new_data["Health"])
naive_interpretation = naive_bayes.perform_methods(data_type)
naive_results = health_query.perform_methods(filename, naive_interpretation, amount)
for i, j in naive_results.iterrows():
st.header(j["Titles"])
st.write(j["Date Published"])
st.write(j["Abstract"])
st.success("Complete!")
if gini_classification:
with st.spinner("Classifying with Gini Index..."):
new_data = decision_tree.import_data(data_type)
st.area_chart(new_data["Health"])
gini_interpretation = decision_tree.perform_gini_index(data_type)
gini_results = health_query.perform_methods(filename, gini_interpretation, amount)
for i, j in gini_results.iterrows():
st.header(j["Titles"])
st.write(j["Date Published"])
st.write(j["Abstract"])
st.success("Complete!")
if entropy_classification:
with st.spinner("Classifying with Entropy..."):
new_data = decision_tree.import_data(data_type)
st.area_chart(new_data["Health"])
entropy_interpretation = decision_tree.perform_entropy(data_type)
entropy_results = health_query.perform_methods(filename, entropy_interpretation, amount)
for i, j in entropy_results.iterrows():
st.header(j["Titles"])
st.write(j["Date Published"])
st.write(j["Abstract"])
st.success("Complete!")
if svm_classification:
with st.spinner("Classifying with Support Vector Machine..."):
new_data = svm.import_data(data_type)
st.area_chart(new_data["Health"])
svm_interpretation = svm.perform_methods(data_type)
svm_results = health_query.perform_methods(filename, svm_interpretation, amount)
for i, j in svm_results.iterrows():
st.header(j["Titles"])
st.write(j["Date Published"])
st.write(j["Abstract"])
st.success("Complete!")
elif choices == 'Visualize':
st.subheader("Visualize")
df = pd.read_csv(data)
all_col = df.columns.to_list()
all_coll = df.columns.to_list()
if st.sidebar.checkbox("Bar Chart"):
bar = st.multiselect("Select Features ", all_col, key='a')
new_df = df[bar]
st.bar_chart(df[bar])
if st.sidebar.checkbox("Area_chart"):
area = st.multiselect("Select Features", all_col)
new_dff = df[area]
st.area_chart(df[area])
if st.sidebar.checkbox("Line Chart"):
st.line_chart(df)
if st.sidebar.checkbox("Plotly charts"):
X1 = st.selectbox("select ", all_col, key='b')
X2 = st.selectbox("select", all_col, key='m')
X3 = st.selectbox("select", all_col)
hist_data = [df[X1], df[X2], df[X3]]
group_labels = [X1, X2, X3]
fig = ff.create_distplot(hist_data,
group_labels,
bin_size=[.1, .25, .5])
st.plotly_chart(fig, use_container_width=True)
def main():
#df = load_data()
data = st.file_uploader("Upload a Dataset", type=["csv", "txt"])
if data is not None:
df = pd.read_csv(data)
page = st.sidebar.selectbox(
"Choose a page",
['Homepage', 'Exploration', 'Plots', 'Prediction'])
if page == 'Homepage':
st.title('Wine Alcohol Class Prediction')
st.text('Select a page in the sidebar')
st.dataframe(df)
elif page == 'Exploration':
st.title('Explore the Wine Data-set')
if st.checkbox("Show Shape"):
st.dataframe(df.shape)
if st.checkbox("Show Columns"):
all_columns = df.columns.to_list()
st.dataframe(all_columns)
if st.checkbox('Show column descriptions'):
st.dataframe(df.describe())
if st.checkbox("Show Selected Columns"):
selected_columns = st.multiselect("Select Columns",
all_columns)
new_df = df[selected_columns]
st.dataframe(new_df)
if st.checkbox("Show Value Counts"):
st.dataframe(df.iloc[:, -1].value_counts())
st.markdown('### Analysing column relations')
st.text('Correlations:')
fig, ax = plt.subplots(figsize=(10, 10))
sns.heatmap(df.corr(), annot=True, ax=ax)
st.pyplot()
st.text('Effect of the different classes')
sns.pairplot(df,
vars=[
'magnesium', 'flavanoids', 'nonflavanoid_phenols',
'proline'
],
hue='alcohol')
st.pyplot()
elif page == 'Plots':
st.subheader("Data Visualization")
st.title('Plots')
if st.checkbox("Show Value Counts"):
st.write(df.iloc[:, -1].value_counts().plot(kind='bar'))
st.pyplot()
# Customizable Plot
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox(
"Select Type of Plot",
["area", "bar", "line", "hist", "box", "kde"])
selected_columns_names = st.multiselect("Select Columns To Plot",
all_columns_names)
if st.button("Generate Plot"):
st.success("Generating Customizable Plot of {} for {}".format(
type_of_plot, selected_columns_names))
# Plot By Streamlit
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
# Custom Plot
elif type_of_plot:
cust_plot = df[selected_columns_names].plot(
kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
else:
st.title('Modelling')
model, accuracy = train_model(df)
st.write('Accuracy: ' + str(accuracy))
st.markdown('### Make prediction')
st.dataframe(df)
row_number = st.number_input('Select row',
min_value=0,
max_value=len(df) - 1,
value=0)
st.markdown('#### Predicted')
st.text(
model.predict(
df.drop(['alcohol'],
axis=1).loc[row_number].values.reshape(1, -1))[0])
```python
import streamlit as st
import numpy as np
import pandas as pd
```
"""
#chart
# 20行3列のランダムな行列を生成
df = pd.DataFrame(np.random.rand(20, 3), columns=['a', 'b', 'c'])
#グラフを表示
st.line_chart(df) #折れ線グラフ
st.area_chart(df) # エリアチャート
st.bar_chart(df) # 棒グラフ
#mapを表示する
df = pd.DataFrame(
np.random.rand(100, 2) / [50, 50] + [35.69, 139.70],
columns=['lat', 'lon'] #lat=緯度, lon=経度
)
# st.write(df)
st.map(df)
#画像を表示する
st.write('Display Image')
def main():
"""Mortality Prediction App"""
st.title("Hepatitis Disease Mortality Prediction App")
# st.markdown(html_temp.format("royalblue"),unsafe_allow_html=True)
menu = ["Home","Login","SignUp"]
submenu = ["Plot","Prediction","Metrics"]
choice = st.sidebar.selectbox("Menu",menu)
if choice == "Home":
st.subheader("Home")
st.text("Welcome")
elif choice == "Login":
username = st.sidebar.text_input("Username")
password = st.sidebar.text_input("Password",type="password")
if st.sidebar.checkbox("Login"):
create_usertable()
hashed_pswd = generate_hashes(password)
result = login_user(username,verify_hashes(password,hashed_pswd))
# if password == "12345":
if result:
st.success("Welcome {}".format(username))
activity = st.selectbox("Activity",submenu)
if activity == "Plot":
st.subheader("Data Visualization Plot")
df = pd.read_csv("data/cleaned_data.csv")
st.dataframe(df)
# freq distribution plot
freq_df = pd.read_csv("data/Age_frq_Dist.csv")
st.bar_chart(freq_df["count"])
st.dataframe(freq_df)
df['class'].value_counts().plot(kind="bar")
st.pyplot()
if st.checkbox("Area Chart"):
all_columns = df.columns.to_list()
feat_choices = st.multiselect("Choose a feature" , all_columns)
new_df = df[feat_choices]
st.area_chart(new_df)
# prediction
elif activity == "Prediction":
st.subheader("Predictive Analytics")
age = st.number_input("Age",7,80)
sex = st.radio("Sex",tuple(gender_dict.keys()))
steroid = st.radio("Do You Take Steroids?",tuple(feature_dict.keys()))
antivirals = st.radio("Do You Take Antivirals?",tuple(feature_dict.keys()))
fatique = st.radio("Do You Have Fatique?",tuple(feature_dict.keys()))
spiders = st.radio("Presence of Spider Naeve",tuple(feature_dict.keys()))
ascites = st.selectbox("Ascites",tuple(feature_dict.keys()))
varices = st.selectbox("Presence of Varices",tuple(feature_dict.keys()))
bilirubin = st.number_input("bilirubin Content",0.0,8.0)
alk_phosphate = st.number_input("Alkaline Phosphate Content",0.0,296.0)
sgot = st.number_input("Sgot",0.0,648.0)
albumin = st.number_input("Albumin",0.0,6.4)
protime = st.number_input("Prothrombin Time",0.0,100.0)
histology = st.radio("Histology",tuple(feature_dict.keys()))
feature_list = [age,get_value(sex,gender_dict),get_fvalue(steroid),get_fvalue(antivirals),get_fvalue(fatique),get_fvalue(spiders)
,get_fvalue(ascites),get_fvalue(varices),bilirubin,alk_phosphate,sgot,albumin,int(protime),get_fvalue(histology)]
st.write(len(feature_list))
st.write(feature_list)
pretty_result = {"age":age,"sex":sex,"steroid":steroid,"antivirals":antivirals,"fatique":fatique,"spiders":spiders,
"ascites":ascites , "varices":varices ,"bilirubin":bilirubin,"alk_phosphate":alk_phosphate,"sgot":sgot,
"albumin":albumin,"protime":protime,"histology":histology}
st.json(pretty_result)
single_sample = np.array(feature_list).reshape(1,-1)
# # Machine Learning models
model_choice = st.selectbox("Select Model",["LR","KNN","DecisionTree"])
if st.button("Predict"):
if model_choice == "KNN":
loaded_model = load_model("models/KNN_HepatitisB_model.pkl")
prediction=loaded_model.predict(single_sample)
pred_prob=loaded_model.predict_proba(single_sample)
elif model_choice == "DecisionTree":
loaded_model=load_model("models/decision_tree_clf_HepatitisB_model.pkl")
prediction=loaded_model.predict(single_sample)
pred_prob=loaded_model.predict_proba(single_sample)
else:
loaded_model = load_model("models/LogisticReg_HepatitisB_model.pkl")
prediction=loaded_model.predict(single_sample)
pred_prob=loaded_model.predict_proba(single_sample)
# st.write(prediction)
# # prediction_label = {"Die":1,"Live":2}
# # final_result = get_key(prediction,prediction_label)
if prediction == 1:
st.warning("Patient Dies")
pred_probability_score = {"Die":pred_prob[0][0]*100 , "Live":pred_prob[0][1]*100}
st.subheader("Prediction Probability Score using {}".format(model_choice))
st.json(pred_probability_score)
st.subheader("Prescriptive Analytics")
st.markdonw(prescriptive_message_temp,unsafe_allow_html=True)
else:
st.success("Patient Lives")
pred_probability_score = {"Die":pred_prob[0][0]*100,"Live":pred_prob[0][1]*100}
st.subheader("Prediction Probability Score using {}".format(model_choice))
st.json(pred_probability_score)
if st.checkbox("Interpret"):
if model_choice == "KNN":
loaded_model = load_model("models/KNN_HepatitisB_model.pkl")
elif model_choice == "DecisionTree":
loaded_model = load_model("models/decision_tree_clf_HepatitisB_model.pkl")
else:
loaded_model = load_model("models/LogisticReg_HepatitisB_model.pkl")
# loaded_model = load_model("models/logistic_regression_model.pkl")
# 1 Die and 2 Live
df = pd.read_csv("data/cleaned_data.csv")
x = df[['age', 'sex', 'steroid', 'antivirals','fatigue','spiders', 'ascites','varices', 'bilirubin', 'alk_phosphate', 'sgot', 'albumin', 'protime','histology']]
feature_names = ['age', 'sex', 'steroid', 'antivirals','fatigue','spiders', 'ascites','varices', 'bilirubin', 'alk_phosphate', 'sgot', 'albumin', 'protime','histology']
class_names = ['Die(1)','Live(2)']
explainer = lime.lime_tabular.LimeTabularExplainer(x.values,feature_names=feature_names, class_names=class_names,discretize_continuous=True)
# The Explainer Instance
exp = explainer.explain_instance(np.array(feature_list), loaded_model.predict_proba,num_features=13, top_labels=1)
exp.show_in_notebook(show_table=True, show_all=False)
# exp.save_to_file('lime_oi.html')
st.write(exp.as_list())
new_exp = exp.as_list()
label_limits = [i[0] for i in new_exp]
# st.write(label_limits)
label_scores = [i[1] for i in new_exp]
plt.barh(label_limits,label_scores)
st.pyplot()
plt.figure(figsize=(20,10))
fig = exp.as_pyplot_figure()
st.pyplot()
else:
st.warning("Incorrect Username/Password")
st.info("Data Visualization")
df = get_dataframe(csv_filename)
if st.checkbox("Show Value Counts"):
fig, ax = plt.subplots()
st.write(df.iloc[:,-1].value_counts().plot(kind='bar'))
st.pyplot(fig)
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox("Select Type of Plot",["area","bar","line","hist","box","kde"])
selected_columns_names = st.multiselect("Select Columns To Plot",all_columns_names)
if st.button("Generate Plot"):
st.success("Generating Customizable Plot of {} for {}".format(type_of_plot,selected_columns_names))
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
if type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
if type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
if type_of_plot == 'hist' or 'box' or 'kde':
if len(selected_columns_names)>1:
st.warning('select one coloumn')
else:
fig, ax = plt.subplots()
cust_plot= df[get_singleton(selected_columns_names)].plot(kind=type_of_plot)
st.write(cust_plot)
# Chart Table
chartTable = pd.DataFrame(
np.random.rand(25, 5),
columns=["A", "B", "X", "Y", "Z"]
)
st.subheader("Chart Table")
st.line_chart(chartTable)
# Colored Chart Table
coloredChartTable = pd.DataFrame(
np.random.rand(25, 3),
columns=["A", "B", "C"]
)
st.subheader("Colored Chart")
st.area_chart(coloredChartTable)
# Bar Chart
barChart = pd.DataFrame(
np.random.rand(25, 3),
columns=["A", "B", "C"]
)
st.subheader("Bar Chart")
st.bar_chart(barChart)
# Displaying Plots
import matplotlib.pyplot as plt
arr = np.random.normal(1, 1, size=100)
fig, ax = plt.subplots()
ax.hist(arr, bins=20)
data = datasets.load_breast_cancer()
st.write("""dataset breast""")
else:
data = datasets.load_wine()
st.write("""dataset wine""")
x = data.data
y = data.target
x1 = pd.DataFrame(x, columns=data.feature_names)
x2 = pd.DataFrame(y)
return x, y, x1, x2
x, y, x1, x2 = select_dataset(dataset)
#x1=frame(dataset)
st.write(""" ## SHAPE OF DATA IS""", x.shape)
st.write(""" ## No of classes """, len(np.unique(y)))
box = st.selectbox(""" display data """, options=["inputdata", "targetdata"])
if box == "inputdata":
st.write(""" ## Input Data""", x1.head())
elif box == "targetdata":
st.write(""" ## target Data""", x2.head())
#st.write("""# heat map """,sn.heatmap(x1.corr(),annot=True))
#box1=st.selectbox("""## PLOT""",options=["inputdata","targetdata"])
st.bar_chart(x1)
st.area_chart(x1)
st.altair_chart(x1)
import bs4
Href = "https://app.powerbi.com/reportEmbed?reportId=55021219-4186-4a3b-b904-4df92fd29bb0&autoAuth=true&ctid=bf93bb5e-ecf0-4e3d-be0e-79b5cc527a48&config=eyJjbHVzdGVyVXJsIjoiaHR0cHM6Ly93YWJpLWluZGlhLWNlbnRyYWwtYS1wcmltYXJ5LXJlZGlyZWN0LmFuYWx5c2lzLndpbmRvd3MubmV0LyJ9"
st.write(Href)
def main():
""" Machine Learning Dataset Explorer"""
st.title("Machine Learning Dataset Explorer")
st.subheader("Simple Data Science Explorer with Streamlit")
html_temp = """
<div style="background-color:tomato;">
<p style="color:white; font-size: 50px">Frase aleatória</p>
<div>
"""
st.markdown(html_temp, unsafe_allow_html=True)
def file_selector(folder_path='.'):
filenames = os.listdir(folder_path)
selected_filename = st.selectbox("Escolhar um arquivo", filenames)
return os.path.join(folder_path, selected_filename)
filename = file_selector()
st.info("Você escolheu {}".format(filename))
#Ler os dados
df = pd.read_csv(filename)
# Mostrar o dataset
if st.checkbox("Mostrar DataSet"):
number = st.number_input("Número de linhas para visualizar", 5, 10)
st.dataframe(df.head(number))
#Mostrar colunas
if st.button("Nomes das Colunas"):
st.write(df.columns)
#Mostrar formatos
if st.checkbox("Formato do Dataset"):
st.write(df.shape)
data_dim = st.radio("Show Dimension By", ("Rows", "Columns"))
if data_dim == 'Columns':
st.text("Número de Colunas")
st.write(df.shape[1])
elif data_dim == "Rows":
st.text("Número de linhas")
st.write(df.shape[0])
else:
st.write(df.shape)
#Escolher colunas
if st.checkbox("Selecione as colunas desejadas"):
all_columns = df.columns.tolist()
selected_columns = st.multiselect("Escolha", all_columns)
new_df = df[selected_columns]
st.dataframe(new_df)
#Mostrar valores
if st.button("Valores"):
st.text("Valores em classes")
st.write(df.iloc[:, 0].value_counts()) #moradores
st.write(df.iloc[:, 1].value_counts()) #idosos
st.write(df.iloc[:, -1].value_counts()) #crianças
st.write(df.iloc[:, -2].value_counts()) #familias
#Mostrar Datatypes
if st.button("DataTypes"):
st.write(df.dtypes)
#Mostrar sumário
if st.checkbox("Sumário"):
st.write(df.describe().T)
#Visualização
st.subheader("Visualização dos dados")
#Corelação
#Seaborn
if st.checkbox("Seaborn Plot"):
st.write(sns.heatmap(df.corr(), annot=True))
st.pyplot
#Count plot
if st.checkbox("Plot of Value Counts"):
st.text("Value Counts By Target")
all_columns_names = df.columns.tolist()
primary_col = st.selectbox("Primary Columm to GroupBy",
all_columns_names)
selected_columns_names = st.multiselect("Select Columns",
all_columns_names)
if st.button("Plot"):
st.text("Generate Plot")
if selected_columns_names:
vc_plot = df.groupby(
primary_col)[selected_columns_names].count()
else:
vc_plot = df.iloc[:, -1].value_counts()
st.write(vc_plot.plot(kind="bar"))
st.pyplot()
#Pie chart
if st.checkbox("Pie Plot"):
all_columns_names = df.columns.tolist()
selected_column = st.selectbox("Selecione a coluna desejada",
all_columns_names)
if st.button("Gerar Pie Plot"):
st.success("Gerando um Pie Plot")
st.write(
df[selected_column].value_counts().plot.pie(autopct="%1.1f%%"))
st.pyplot()
#Plot customizado
st.subheader("Plot Customizado")
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox("Selecione o tipo de plot",
['area', 'bar', 'line', 'hist', 'box', 'kde'])
selected_columns_names = st.multiselect("Selecione as colunas",
all_columns_names)
if st.button("Gerar Plot"):
st.success("Gerando plot de {} para {}".format(type_of_plot,
selected_columns_names))
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
elif type_of_plot:
cust_plot = df[selected_columns_names].plot(kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
def main():
"""Automated ML App"""
#st.title('Machine Learning Application')
activities = ["Home", "EDA", "Plots", "ML_Algorithms", "Neural Network"]
choice = st.sidebar.selectbox("Menu", activities)
html_temp = """
<div
style="background-color:royalblue;padding:10px;border-radius:10px">
<h1 style="color:white;text-align:center;font-style: italic;">Classifying the survival of patients with heart failure using Various Machine Learning Algorithms</h1>
</div>
"""
components.html(html_temp)
#data = st.file_uploader("Upload a Dataset", type=["csv","txt","xlsx"])
data = pd.read_csv('heart_failure.csv')
if choice == 'EDA':
st.subheader("Exploratory Data Analysis using Pandas Profiling")
if data is not None:
df = pd.read_csv('heart_failure.csv')
st.dataframe(df.head())
lable = preprocessing.LabelEncoder()
for col in df.columns:
df[col] = lable.fit_transform(df[col])
#pandas profiling
profile = ProfileReport(df)
st_profile_report(profile)
elif choice == 'Plots':
st.subheader("Data Visualization")
if data is not None:
df = pd.read_csv('heart_failure.csv')
st.dataframe(df.head())
lable = preprocessing.LabelEncoder()
for col in df.columns:
df[col] = lable.fit_transform(df[col])
if st.checkbox("Show Value Counts"):
st.write(df.iloc[:, -1].value_counts().plot(kind='bar'))
st.pyplot()
#Customized Plot
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox(
"Select Type of Plot",
["area", "bar", "line", "hist", "box", "kde"])
selected_columns_names = st.multiselect("Select Columns To Plot",
all_columns_names)
if st.button("Generate Plot"):
st.success("Generating Customizable Plot of {} for {}".format(
type_of_plot, selected_columns_names))
# Plot By Streamlit
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
# Custom Plot
elif type_of_plot:
cust_plot = df[selected_columns_names].plot(kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
elif choice == 'ML_Algorithms':
st.subheader("Machine Learning Algorithms")
if data is not None:
df = pd.read_csv('heart_failure.csv')
st.dataframe(df.head())
lable = preprocessing.LabelEncoder()
for col in df.columns:
df[col] = lable.fit_transform(df[col])
if st.checkbox("Summary"):
st.write(df.describe())
X = df.iloc[:, :-1].values
y = df.iloc[:, -1].values
#col_name = st.selectbox("Select Column Name",["X","y"])
#if col_name == 'X':
# st.dataframe(X)
#elif col_name == 'y':
# st.dataframe(y)
st.write("Number of classes", len(np.unique(y)))
params = dict()
classifer_name = st.sidebar.selectbox(
"Select Classifer",
("SVM Linear", "SVM Radial", "Decision Tree", "Random Forest"))
#add parameters
def add_parameters(clf_name):
"""Selection of parameters"""
if clf_name == "SVM Linear":
C = st.sidebar.slider("C", 0.01, 15.0)
params["C"] = C
elif clf_name == "SVM Radial":
C = st.sidebar.slider("C", 0.01, 15.0)
params["C"] = C
elif clf_name == "Decision Tree":
max_depth = st.sidebar.slider("max_depth", 2, 15)
max_leaf_nodes = st.sidebar.slider("max_leaf_nodes", 2, 20)
params["max_depth"] = max_depth
params["max_leaf_nodes"] = max_leaf_nodes
elif clf_name == "Random Forest":
max_depth = st.sidebar.slider("max_depth", 2, 15)
n_estimators = st.sidebar.slider("n_estimators", 1, 200)
params["max_depth"] = max_depth
params["n_estimators"] = n_estimators
return params
add_parameters(classifer_name)
#get classifers
def get_classifiers(clf_name, params):
clf = None
if clf_name == "SVM Linear":
clf = SVC(C=params["C"], kernel='linear')
elif clf_name == "SVM Radial":
clf = SVC(C=params["C"], kernel='rbf')
elif clf_name == "Decision Tree":
clf = DecisionTreeClassifier(
max_depth=params["max_depth"],
max_leaf_nodes=params["max_leaf_nodes"],
random_state=100)
elif clf_name == "Random Forest":
clf = RandomForestClassifier(
n_estimators=params["n_estimators"],
max_depth=params["max_depth"],
random_state=100)
return clf
clf = get_classifiers(classifer_name, params)
#Classification
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.2,
random_state=100)
clf.fit(X_train, y_train)
y_pred = clf.predict(X_test)
acc = accuracy_score(y_test, y_pred)
st.write(
f'<div style="color: #1C2331; font-size: medium; font-style: italic; padding: 15px; background-color:#b2dfdb;border-radius:5px;">Classifier = {classifer_name}</div></br>',
unsafe_allow_html=True)
clf_report = classification_report(y_test, y_pred)
st.success(f"Classification Report:\n\n {clf_report}")
st.warning(f"accuracy = {acc}")
for i in range(1, 10):
st.write("Actual=%s, Predicted=%s" % (y_test[i], y_pred[i]))
elif choice == 'Neural Network':
st.subheader("Neural Networks (MLPClassifier)")
if data is not None:
df = pd.read_csv('heart_failure.csv')
st.dataframe(df.head())
lable = preprocessing.LabelEncoder()
for col in df.columns:
df[col] = lable.fit_transform(df[col])
X = df.iloc[:, :-1].values
y = df.iloc[:, -1].values
params = dict()
classifer_name = "MLPClassifier"
def add_parameters(clf_name):
"""Selection of parameters"""
if clf_name == "MLPClassifier":
max_iter = st.sidebar.slider("max_iter", 2, 30)
params["max_iter"] = max_iter
return params
add_parameters(classifer_name)
#get classifers
def get_classifiers(clf_name, params):
clf = None
if clf_name == "MLPClassifier":
clf = MLPClassifier(max_iter=params["max_iter"])
return clf
clf = get_classifiers(classifer_name, params)
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.2,
random_state=100)
clf.fit(X_train, y_train)
y_pred = clf.predict(X_test)
st.write(
f'<div style="color: #1C2331; font-size: medium; font-style: italic; padding: 15px; background-color:#b2dfdb;border-radius:5px;">Classifier = {classifer_name}</div></br>',
unsafe_allow_html=True)
clf_report = classification_report(y_test, y_pred)
st.success(f"Classification Report:\n\n {clf_report}")
acc = accuracy_score(y_test, y_pred)
st.warning(f"accuracy = {acc}")
for i in range(1, 10):
st.write("Actual=%s, Predicted=%s" % (y_test[i], y_pred[i]))
def data_display():
"# In this section we will try to explore ways to display our data"
# Let’s add a title to test things out
st.title("My Cool Streamlit App!!!")
# Let's write something
# write any text
st.write("Here's our first attempt at using data to create a table:")
# write dataframes
st.write(
pd.DataFrame({"first column": [1, 2, 3, 4], "second column": [10, 20, 30, 40]})
)
# write JSON
st.write({"Name": "John", "Country": "USA"})
# Emojis are love!
st.write("Display some cool emojis :sunglasses:")
# st.write accepts chart objects too!
df2 = pd.DataFrame(np.random.randn(200, 3), columns=["a", "b", "c"])
c = (
alt.Chart(df2)
.mark_circle()
.encode(x="a", y="b", size="c", color="c", tooltip=["a", "b", "c"])
)
# st.write(c)
# you can also use
st.text("you can also use st.text or st.markdown as well")
# let me showcase some Magic!
"""
# My Cool Streamlit App!!!
Here's our first attempt at using data to create a table:
"""
df = pd.DataFrame({"first column": [1, 2, 3, 4], "second column": [10, 20, 30, 40]})
# df
# you can also use st.dataframe and st.table
"Display a df with `st.dataframe(df)` and `st.table(df)`"
st.dataframe(df)
st.table(df)
# let's display some charts
"let's display charts"
chart_data = pd.DataFrame(np.random.randn(50, 3), columns=["a", "b", "c"])
# line_chart
"Display a line chart figure with `st.line_chart`"
st.line_chart(chart_data)
# area_chart
"Display a area chart figure with `st.area_chart`"
st.area_chart(chart_data)
# bar_chart
"Display a bar chart figure with `st.bar_chart`"
st.bar_chart(chart_data)
# let's draw Map chart
map_data = pd.DataFrame(
np.random.randn(1000, 2) / [50, 50] + [37.76, -122.4], columns=["lat", "lon"]
)
"Display a map figure with `st.map`"
st.map(map_data)
"Display a matplotlib.pyplot figure with `st.pyplot`"
arr = np.random.normal(1, 1, size=100)
fig, ax = plt.subplots()
ax.hist(arr, bins=20)
st.pyplot(fig)
"Display a chart using the Altair library with `st.altair_chart`"
st.altair_chart(c, use_container_width=True)
"Display an interactive Plotly chart using `st.plotly_chart`"
# Add histogram data
x1 = np.random.randn(200) - 2
x2 = np.random.randn(200)
x3 = np.random.randn(200) + 2
# Group data together
hist_data = [x1, x2, x3]
group_labels = ["Group 1", "Group 2", "Group 3"]
# Create distplot with custom bin_size
fig = ff.create_distplot(hist_data, group_labels, bin_size=[0.1, 0.25, 0.5])
st.plotly_chart(fig, use_container_width=True)
"""
import streamlit as st import pandas as pd import numpy as np chart_data = pd.DataFrame(np.random.randn(20, 3), columns=['a', 'b', 'c']) st.area_chart(chart_data)
])
st.subheader('Total Energy Generation in ' + country + ' (MW)')
forecast_horizon = st.sidebar.slider(label='Forecast Horizon (hours)',
min_value=12,
max_value=168,
value=48)
window_length = st.sidebar.slider(label='Window Length', min_value=1, value=30)
country_code = COUNTRY_MAPPINGS[country]
df = get_energy_data(country_code)
#Plotting total energy generation for selected country
st.area_chart(df, use_container_width=False, width=800)
cols_renewable = ['Wind Onshore', 'Wind Offshore', 'Solar']
#Selecting the renewable energy columns,
#Only if they are available in the dataframe
df = df[df.columns & cols_renewable]
for item in df.columns:
smape = calculate_smape(df[[item]], regressor, forecast_horizon,
window_length)
st.subheader(item + ' Energy Generation Forecast in ' + country + ' (MW)')
#Generating and plotting a forecast for each renewable energy type
df_forecast = generate_forecast(df[[item]], regressor, forecast_horizon,
window_length)
def main():
"""Semi Automated ML App With Streamlit """
activities = ["Exploratory Data Analysis", "Plots"]
choice = st.sidebar.selectbox("Select Activities", activities)
if choice == 'Exploratory Data Analysis':
st.subheader("Exploratory Data Analysis")
data = st.file_uploader("Upload a Dataset of Your Choice",
type=["csv", "txt"])
if data is not None:
df = pd.read_csv(data)
st.dataframe(df.head())
if st.checkbox("Show Shape"):
st.write(df.shape)
if st.checkbox("Show Columns"):
all_columns = df.columns.to_list()
st.write(all_columns)
if st.checkbox("Summary"):
st.write(df.describe())
if st.checkbox("Show Selected Columns"):
selected_columns = st.multiselect("Select Columns",
all_columns)
new_df = df[selected_columns]
st.dataframe(new_df)
if st.checkbox("Show Value Counts"):
st.dataframe(df.iloc[:, -1].value_counts())
if st.checkbox("Correlation Plot(Matplotlib)"):
plt.matshow(df.corr())
st.pyplot()
if st.checkbox("Correlation Plot(Seaborn)"):
st.write(sns.heatmap(df.corr(), annot=True))
st.pyplot()
# if st.checkbox("Pie Plot"):
# all_columns = df.columns.to_list()
# column_to_plot = st.selectbox("Select 1 Column", all_columns)
# pie_plot = df[column_to_plot].value_counts().plot.pie(autopct="%1.1f%%")
# st.write(pie_plot)
# st.pyplot()
if st.checkbox("Pie plot"):
all_columns = df.columns.to_list()
column_to_plot = st.selectbox("Select Column", all_columns)
pie_plot = df[column_to_plot].value_counts().plot.pie(
autopct="%1.1f%%")
st.write(pie_plot)
st.pyplot()
elif choice == 'Plots':
st.subheader("Data Visualization")
data = st.file_uploader("Upload a Dataset",
type=["csv", "txt", "xlsx"])
if data is not None:
df = pd.read_csv(data)
st.dataframe(df.head())
if st.checkbox("Show Value Counts"):
st.write(df.iloc[:, -1].value_counts().plot(kind='bar'))
st.pyplot()
# Customizable Plot
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox(
"Select Type of Plot",
["area", "bar", "line", "hist", "box", "kde"])
selected_columns_names = st.multiselect("Select Columns To Plot",
all_columns_names)
if st.button("Generate Plot"):
st.success("Generating Customizable Plot of {} for {}".format(
type_of_plot, selected_columns_names))
# Plot By Streamlit
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
# Custom Plot
elif type_of_plot:
cust_plot = df[selected_columns_names].plot(
kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
def main():
html_page = """
<div style="text-align: center;">
<div>
<h1>Machine Learning | Pedictive Analysis | Data Analytics</h1>
</div>
<div>
<h2>Drag and Drop any dataset and predict or analyse</h2>
</div>
</div>
"""
st.markdown(html_page, unsafe_allow_html=True)
activities = [
"Exploratory Data Analysis", "Plot", "Model Building", "About"
]
choice = st.sidebar.selectbox("Select Activities", activities)
if choice == "Exploratory Data Analysis":
st.subheader("Exploratory Data Analysis")
# Drag and Drop Feature
data = st.file_uploader("Upload Dataset", type=["csv", "txt"])
if data is not None:
df = pd.read_csv(data)
st.dataframe(df.head())
if st.checkbox("Show Shape of the dataset"):
st.write(df.shape)
if st.checkbox("Show Columns of the dataset"):
all_columns = df.columns.to_list()
st.write(all_columns)
if st.checkbox("Select Columns to Show"):
all_columns = df.columns.to_list(
) # doing this again so that we can use this functionality without depending on above
selected_columns = st.multiselect("Select the Columns",
all_columns)
new_df = df[selected_columns]
st.dataframe(new_df)
if st.checkbox("Show Summary of the dataset"):
st.write(df.describe())
if st.checkbox("Show Value Counts"):
st.write(df.iloc[:, -1].value_counts())
elif choice == "Plot":
st.subheader("Data Visualization")
data = st.file_uploader("Upload Dataset", type=["csv", "txt"])
if data is not None:
df = pd.read_csv(data)
st.dataframe(df.head())
if st.checkbox("Correlation Map with Seaborn"):
st.write(sns.heatmap(df.corr(), annot=True))
st.pyplot()
if st.checkbox("Pie Chart"):
all_columns = df.columns.to_list()
columns_to_plot = st.selectbox("Select 1 Column", all_columns)
pie_plot = df[columns_to_plot].value_counts().plot.pie(
autopct="%1.1f%%")
st.write(pie_plot)
st.pyplot()
all_columns_names = df.columns.to_list()
type_of_plot = st.selectbox(
"Select Type of Plot",
["area", "bar", "line", "hist", "box", "kde"])
selected_columns_names = st.multiselect("Select Columns To Plot",
all_columns_names)
if st.button("Generate Plot"):
st.success("Generating Customizable Plot of {} for {}".format(
type_of_plot, selected_columns_names))
# Plots By Streamlit
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
# custom plot
elif type_of_plot:
cust_plot = df[selected_columns_names].plot(kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
elif choice == "Model Building":
st.subheader("Building Machine Learning Model")
data = st.file_uploader("Upload Dataset", type=["csv", "txt"])
if data is not None:
df = pd.read_csv(data)
st.dataframe(df.head())
# Model Building
X = df.iloc[:, 0:-1]
Y = df.iloc[:, -1]
seed = 7
# model
models = []
models.append(("LR", LogisticRegression()))
models.append(("LDA", LinearDiscriminantAnalysis()))
models.append(("KNN", KNeighborsClassifier()))
models.append(("CART", DecisionTreeClassifier()))
models.append(("NB", GaussianNB()))
models.append(("SVM", SVC()))
# Evaluate each model one by one
# List
model_names = []
model_mean = []
model_std = []
all_models = []
scoring = 'accuracy'
for name, model in models:
kfold = model_selection.KFold(n_splits=10, random_state=seed)
cv_results = model_selection.cross_val_score(model,
X,
Y,
cv=kfold,
scoring=scoring)
model_names.append(name)
model_mean.append(cv_results.mean())
model_std.append(cv_results.std())
accuracy_result = {
"model_name": name,
"model_accuracy": cv_results.mean(),
"Standard_deviation": cv_results.std()
}
all_models.append(accuracy_result)
if st.checkbox("Metrics as Table"):
st.dataframe(
pd.DataFrame(zip(model_names, model_mean, model_std),
columns=[
"Model Name", "Model Accuracy",
"Standard Deviation"
]))
if st.checkbox("Metrics as JSON"):
st.json(all_models)
elif choice == "About":
st.subheader("About Developer")
html_page = """
<div style = "background-color:tomato; padding:50px">
<p style="font-size:25px"> SHAIL MODI <br> B. Tech - Computer Engineering <br> ✉ [email protected] <br> K.J.Somaiya College Of Engineering </p>
<p>© shailmodi<p>
</div>
"""
st.markdown(html_page, unsafe_allow_html=True)
# Group
if st.checkbox("Show Bar Chart Plot"):
v_group = data.groupby('species')
st.bar_chart(v_group)
st.pyplot()
# Line
if st.checkbox("Show Line Plot"):
v_group = data.groupby('species')
st.bar_chart(v_group)
st.pyplot()
# Area
if st.checkbox("Show Area Plot"):
v_group = data.groupby('species')
st.area_chart(v_group)
st.pyplot()
# Images
@st.cache
def load_image(img):
im = Image.open(os.path.join(img))
return im
species_type = st.radio("Select Species Type",
("setosa", "virginica", "versicolor"))
if species_type == 'setosa':
st.text("Showing Setosa Species")
st.image(load_image('imgs/iris_setosa.jpg'))
def main():
"""Common ML Data Explorer """
st.title("Common ML Data Explorer")
st.subheader("Simple ML App with Streamlit")
img_list = glob.glob("images/*.png")
# st.write(img_list)
# for i in img_list:
# c_image = Image.open(i)
# st.image(i)
all_image = [Image.open(i) for i in img_list]
st.image(all_image)
def file_selector(folder_path='./datasets'):
filenames = os.listdir(folder_path)
selected_filename = st.selectbox('Select a file', filenames)
return os.path.join(folder_path, selected_filename)
filename = file_selector()
st.write('You selected `%s`' % filename)
df = pd.read_csv(filename)
if st.checkbox("Show DataSet"):
number = st.number_input("Number of Rows to View")
st.dataframe(df.head(number))
if st.button("Columns Names"):
st.write(df.columns)
if st.checkbox("Shape of Dataset"):
st.write(df.shape)
data_dim = st.radio("Show Dimension by", ("Rows", "Columns"))
if data_dim == 'Rows':
st.text("Number of Rows")
st.write(df.shape[0])
elif data_dim == 'Columns':
st.text("Number of Columns")
st.write(df.shape[1])
if st.checkbox("Select Columns To Show"):
all_columns = df.columns.tolist()
selected_columns = st.multiselect('Select', all_columns)
new_df = df[selected_columns]
st.dataframe(new_df)
if st.button("Data Types"):
st.write(df.dtypes)
if st.button("Value Counts"):
st.text("Value Counts By Target/Class")
st.write(df.iloc[:, -1].value_counts())
if st.checkbox("Summary"):
st.write(df.describe())
st.subheader("Data Visualization")
# Show Correlation Plots
# Matplotlib Plot
if st.checkbox("Correlation Plot [Matplotlib]"):
plt.matshow(df.corr())
st.pyplot()
# Seaborn Plot
if st.checkbox("Correlation Plot with Annotation[Seaborn]"):
st.write(sns.heatmap(df.corr(), annot=True))
st.pyplot()
# Counts Plots
if st.checkbox("Plot of Value Counts"):
st.text("Value Counts By Target/Class")
all_columns_names = df.columns.tolist()
primary_col = st.selectbox('Select Primary Column To Group By',
all_columns_names)
selected_column_names = st.multiselect('Select Columns',
all_columns_names)
if st.button("Plot"):
st.text("Generating Plot for: {} and {}".format(
primary_col, selected_column_names))
if selected_column_names:
vc_plot = df.groupby(
primary_col)[selected_column_names].count()
else:
vc_plot = df.iloc[:, -1].value_counts()
st.write(vc_plot.plot(kind='bar'))
st.pyplot()
if st.checkbox("Pie Plot"):
all_columns_names = df.columns.tolist()
# st.info("Please Choose Target Column")
# int_column = st.selectbox('Select Int Columns For Pie Plot',all_columns_names)
if st.button("Generate Pie Plot"):
# cust_values = df[int_column].value_counts()
# st.write(cust_values.plot.pie(autopct="%1.1f%%"))
st.write(df.iloc[:, -1].value_counts().plot.pie(autopct="%1.1f%%"))
st.pyplot()
if st.checkbox("BarH Plot"):
all_columns_names = df.columns.tolist()
st.info("Please Choose the X and Y Column")
x_column = st.selectbox('Select X Columns For Barh Plot',
all_columns_names)
y_column = st.selectbox('Select Y Columns For Barh Plot',
all_columns_names)
barh_plot = df.plot.barh(x=x_column, y=y_column, figsize=(10, 10))
if st.button("Generate Barh Plot"):
st.write(barh_plot)
st.pyplot()
st.subheader("Customizable Plots")
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox("Select the Type of Plot",
["area", "bar", "line", "hist", "box", "kde"])
selected_column_names = st.multiselect('Select Columns To Plot',
all_columns_names)
# plot_fig_height = st.number_input("Choose Fig Size For Height",10,50)
# plot_fig_width = st.number_input("Choose Fig Size For Width",10,50)
# plot_fig_size =(plot_fig_height,plot_fig_width)
cust_target = df.iloc[:, -1].name
if st.button("Generate Plot"):
st.success("Generating A Customizable Plot of: {} for :: {}".format(
type_of_plot, selected_column_names))
# Plot By Streamlit
if type_of_plot == 'area':
cust_data = df[selected_column_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_column_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_column_names]
st.line_chart(cust_data)
# Plot By Matplotlib
# elif type_of_plot == 'pie':
# custom_plot = df[selected_column_names].plot(subplots=True,kind=type_of_plot)
# st.write(custom_plot)
# st.pyplot()
elif type_of_plot == 'hist':
custom_plot = df[selected_column_names].plot(kind=type_of_plot,
bins=2)
st.write(custom_plot)
st.pyplot()
elif type_of_plot == 'box':
custom_plot = df[selected_column_names].plot(kind=type_of_plot)
st.write(custom_plot)
st.pyplot()
elif type_of_plot == 'kde':
custom_plot = df[selected_column_names].plot(kind=type_of_plot)
st.write(custom_plot)
st.pyplot()
else:
cust_plot = df[selected_column_names].plot(kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
html_temp = """
<div style="background-color:powderblue;"><p style="color:blue;font-size:60px;"> Hello world colored</p></div>
"""
# html_temp2 = """
# <body style="background-color:red;">
# <p style="color:blue">Hello World Streamlit</p>
# <form>
# <input type="text"/>
# </form>
# </body>
# </html>"""
st.markdown(html_temp, unsafe_allow_html=True)
st.subheader("Feature Engineering and ML Aspect")
if st.checkbox("Show Features"):
all_features = df.iloc[:, 0:-1]
st.text('Features Names:: {}'.format(all_features.columns[0:-1]))
st.dataframe(all_features.head(10))
if st.checkbox("Show Target"):
all_target = df.iloc[:, -1]
st.text('Target/Class Name:: {}'.format(all_target.name))
st.dataframe(all_target.head(10))
all_ml_dict = {
'LR': LogisticRegression(),
'LDA': LinearDiscriminantAnalysis(),
'KNN': KNeighborsClassifier(),
'CART': DecisionTreeClassifier(),
'NB': GaussianNB(),
'SVM': SVC()
}
# models = []
model_choice = st.multiselect('Model Choices', list(all_ml_dict.keys()))
for key in all_ml_dict:
if 'LDA' in key:
st.write(key)
# results = []
# names = []
# allmodels = []
# scoring = 'accuracy'
# for name, model in models:
# kfold = model_selection.KFold(n_splits=10, random_state=seed)
# cv_results = model_selection.cross_val_score(model, X, Y, cv=kfold, scoring=scoring)
# results.append(cv_results)
# names.append(name)
# msg = "%s: %f (%f)" % (name, cv_results.mean(), cv_results.std())
# allmodels.append(msg)
# model_results = results
# model_names = names
# Make Downloadable file as zip,since markdown strips to html
st.markdown("""[google.com](iris.zip)""")
st.markdown("""[google.com](./iris.zip)""")
# def make_zip(data):
# output_filename = '{}_archived'.format(data)
# return shutil.make_archive(output_filename,"zip",os.path.join("downloadfiles"))
def makezipfile(data):
output_filename = '{}_zipped.zip'.format(data)
with ZipFile(output_filename, "w") as z:
z.write(data)
return output_filename
if st.button("Download File"):
DOWNLOAD_TPL = f'[{filename}]({makezipfile(filename)})'
# st.text(DOWNLOAD_TPL)
st.text(DOWNLOAD_TPL)
st.markdown(DOWNLOAD_TPL)
def on_train_begin(self, logs=None):
st.header("Summary")
self._summary_chart = st.area_chart()
self._summary_stats = st.text("%8s : 0" % "epoch")
st.header("Training Log")
st.write(Final_Data)
if st.checkbox("Show all the column Names"):
st.write(Final_Data.columns)
if st.checkbox("Show size of dataset"):
if st.checkbox("Show row size"):
st.write(Final_Data.shape[0])
if st.checkbox("Show column size"):
st.write(Final_Data.shape[1])
if st.checkbox("Show complete dataset size"):
st.write(Final_Data.shape)
if st.checkbox("Show desc of Ratings in final data"):
Final_Data.describe()["Ratings"]
st.write("**displaying final dataset header lines using area chart**")
st.area_chart(Final_Data)
print("Number of NaN values = " + str(Final_Data.isnull().sum()))
duplicates = Final_Data.duplicated(["MovieID", "CustID", "Ratings"])
print("Number of duplicate rows = " + str(duplicates.sum()))
if st.checkbox("Show unique customer & movieId in Total Data:"):
st.write("Total number of movie ratings = ", str(Final_Data.shape[0]))
st.write("Number of unique users = ",
str(len(np.unique(Final_Data["CustID"]))))
st.write("Number of unique movies = ",
str(len(np.unique(Final_Data["MovieID"]))))
if not os.path.isfile("Data/TrainData.pkl"):
Final_Data.iloc[:int(Final_Data.shape[0] *
def main():
def get_table_download_link2(df1, df2, frase):
xlsx_io = io.BytesIO()
writer = pd.ExcelWriter(xlsx_io, engine='xlsxwriter')
df1.to_excel(writer, 'Обучающая выборка')
df2.to_excel(writer, 'Тестовая ваборка')
writer.save()
xlsx_io.seek(0)
media_type = 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'
data = base64.b64encode(xlsx_io.read()).decode("utf-8")
href = f'<a href="data:{media_type};base64,{data}" download="RegressionResults.xlsx" >{frase}</a> (right-click and save)'
st.markdown(href, unsafe_allow_html=True)
def save_train_test_data_xlsx(trainX, trainY, testX, testY, predictions):
cols = []
for i in all_clummn_names:
cols.append(str(i))
cols.remove(selected_y[-1])
cols.append(str(selected_y[-1]) + " ( Selected y )")
df1 = pd.DataFrame(np.concatenate((trainX, trainY), axis=1),
columns=cols)
st.text("Обучающая выборка")
st.write(df1)
cols.pop()
cols.append('Predictions')
cols.append(str(selected_y[-1]) + " ( Real y value )")
predictions = np.reshape(predictions, (predictions.shape[0], 1))
df2 = pd.DataFrame(np.concatenate((np.concatenate(
(testX, predictions), axis=1), testY),
axis=1),
columns=cols)
st.text("Тестовая выборка")
st.write(df2)
get_table_download_link2(df1, df2, "Сохранить результаты xlsx File")
def save_exel(df, frase):
xlsx_io = io.BytesIO()
writer = pd.ExcelWriter(xlsx_io, engine='xlsxwriter')
df.to_excel(writer, 'Результаты кластеризации')
writer.save()
xlsx_io.seek(0)
media_type = 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'
data = base64.b64encode(xlsx_io.read()).decode("utf-8")
href = f'<a href="data:{media_type};base64,{data}" download="ClusteringResults.xlsx" >{frase}</a> (right-click and save)'
st.markdown(href, unsafe_allow_html=True)
def get_file():
file = st.file_uploader(
" Нажмите browse files, чтобы загрузить файл в формате .csv",
type="csv")
show_file = st.empty()
if not file:
show_file.info(
" Файл не загружен. Загрузите файл для анализа в формате .csv")
return
else:
show_file.info(" Загрузка файла выполнена успешно")
return file
def regression_errors_values(testY, predict, n_predict):
st.write("Средняя абсолютная ошибка:",
mean_absolute_error(testY, predict))
st.write("Средняя квадратичная ошибка:",
mean_squared_error(testY, predict))
st.write("Средняя абсолютная ошибка (c нормализацией):",
mean_absolute_error(testY, n_predict))
st.write("Средняя квадратичная ошибка (c нормализацией):",
mean_squared_error(testY, n_predict))
def regression_plot_show(trainX, testY, predict, n_predict, reg_type,
target_label):
xx = [i for i in range(trainX.shape[0])]
plt.figure()
plt.plot(xx[0:testY.size],
testY[0:testY.size],
'o',
color='r',
label='y')
plt.plot(xx[0:testY.size],
predict[0:testY.size],
color='b',
linewidth=2,
label='predicted y')
plt.plot(xx[0:testY.size],
n_predict[0:testY.size],
color='k',
linewidth=2,
label='predicted y with normalize')
plt.ylabel(target_label)
plt.xlabel('Line number in dataset')
plt.legend(loc=4)
plt.title(reg_type)
st.pyplot(plt.show())
def model_linear_Regression(trainX, trainY, testX, testY):
model = LinearRegression(normalize=False)
model.fit(trainX, trainY)
predict = model.predict(testX)
return predict
def model_ridge_Regression(trainX, trainY, testX, testY):
model = Ridge(normalize=True)
model.fit(trainX, trainY)
predict = model.predict(testX)
return predict
def model_lasso(trainX, trainY, testX, testY):
model = Lasso(normalize=False)
model.fit(trainX, trainY)
predict = model.predict(testX)
return predict
def model_random_forest(trainX, trainY, testX, testY):
model = RandomForestRegressor(criterion="mae", bootstrap=True)
model.fit(trainX, trainY)
predict = model.predict(testX)
return predict
st.title("Анализ данных & Машинное обучение")
st.subheader("Загрузите файл с данными для анализа")
file = get_file()
if file:
data = pd.read_csv(file, sep=',')
if st.checkbox("Показать данные"):
num = st.slider("Колличество строк для отображеия ", 5,
data.shape[0])
st.dataframe(data.head(num))
if st.checkbox("Заголовки столбцов"):
st.write(data.columns)
if st.checkbox("Показать размерность"):
data_dm = st.radio("Размерность по", ("Строкам", "Столбцам"))
if data_dm == "Столбцам":
st.write("Число столбцов: " + str(data.shape[1]))
elif data_dm == "Строкам":
st.write("Число строк: " + str(data.shape[0]))
if st.checkbox("Типы данных"):
st.write(data.dtypes)
if st.checkbox("Статистика по значениям"):
st.write(data.describe().T)
if st.checkbox("Рассмотреть отдельные столбцы"):
all_data = data.columns.tolist()
sltd_columns = st.multiselect("Select", all_data)
new_data = data[sltd_columns]
data = new_data
st.dataframe(new_data)
st.header("Визуализация данных")
if st.checkbox("Построить корреляционную матрицу"):
plt.figure(figsize=(10, 10))
plt.title('Correlation between different fearures')
st.write(
sns.heatmap(data.corr(),
vmax=1,
square=True,
annot=True,
cmap='gray_r'))
st.pyplot()
st.subheader("Построение графиков")
all_clummn_names = data.columns.tolist()
plot_type = st.selectbox("Выберите тип графика",
["area", "bar", "line", "hist", "box", "kde"])
selected_columns = st.multiselect(
"Выберите столбцы для построения графика", all_clummn_names)
if st.button("Построить график"):
st.success("Построение графика {} для {}".format(
plot_type, selected_columns))
if plot_type == 'area':
plot_data = data[selected_columns]
st.area_chart(plot_data)
elif plot_type == 'bar':
plot_data = data[selected_columns]
st.bar_chart(plot_data)
elif plot_type == 'line':
plot_data = data[selected_columns]
st.line_chart(plot_data)
elif plot_type:
cust_data = data[selected_columns].plot(kind=plot_type)
st.write(cust_data)
st.pyplot()
if st.checkbox("Круговая диаграмма"):
all_clummn_names = data.columns.tolist()
if st.button("Построить круговую диаграмму"):
st.success("Поостроение круговой диаграммы")
st.write(
data.iloc[:,
-1].value_counts().plot.pie(autopct="%1.1f%%"))
st.pyplot()
if st.checkbox("PCA проекция"):
scaler = StandardScaler()
pca = decomposition.PCA(n_components=2)
X_reduced = pca.fit_transform(scaler.fit_transform(data))
plt.figure(figsize=(6, 6))
plt.scatter(X_reduced[:, 0],
X_reduced[:, 1],
edgecolor='none',
alpha=0.7,
s=40,
cmap=plt.cm.get_cmap('nipy_spectral', 10))
plt.title('Feautures PCA projection')
st.pyplot(plt.show())
if st.checkbox("TSNE проекция"):
scaler = StandardScaler()
tsne = TSNE(random_state=17)
tsne_representation = tsne.fit_transform(
scaler.fit_transform(data))
plt.figure(figsize=(6, 6))
plt.scatter(tsne_representation[:, 0],
tsne_representation[:, 1],
edgecolor='none',
alpha=0.7,
s=40,
cmap=plt.cm.get_cmap('nipy_spectral', 10))
plt.title('Feautures T-sne projection ')
st.pyplot(plt.show())
st.subheader("Применение методов машинного обучения")
if st.checkbox("Выбрать тип решаемой задачи"):
problem_type = st.radio(
"Выберите тип задачи",
("Регрессия", "Кластеризация", "Классификация"))
if problem_type == "Регрессия":
st.write("Выберите столбец у для задачи регрессии")
selected_y = st.multiselect("", all_clummn_names)
reg_type = st.radio("Выберите алгоритм решения",
("Все", "Линейная регрессия", "Ridge",
"Lasso", "Случайный лес"))
test_size_slider = st.slider(
"Выберите размер тестовой выборки %", 1, 100)
if st.checkbox("Выполнить"):
if (selected_y):
st.success(
"Столбец {} выбран успешно".format(selected_y))
train, test = train_test_split(
data, test_size=test_size_slider / 100)
#st.write(test_size_slider/100)
trainX = np.array(train.drop(selected_y, 1))
trainY = np.array(train[selected_y])
testX = np.array(test.drop(selected_y, 1))
testY = np.array(test[selected_y])
st.write(selected_y)
# st.write(trainY.reshape(trainY.shape[0]))
if reg_type == "Все":
reg_type = "Линейная регрессия"
n_predictions = model_linear_Regression(
preprocessing.normalize(trainX), trainY,
preprocessing.normalize(testX), testY)
predictions = model_linear_Regression(
trainX, trainY, testX, testY)
regression_plot_show(trainX, testY, predictions,
n_predictions, reg_type,
selected_y)
regression_errors_values(testY, predictions,
n_predictions)
save_train_test_data_xlsx(trainX, trainY, testX,
testY, predictions)
reg_type = "Ridge"
n_predictions = model_ridge_Regression(
preprocessing.normalize(trainX), trainY,
preprocessing.normalize(testX), testY)
predictions = model_ridge_Regression(
trainX, trainY, testX, testY)
regression_plot_show(trainX, testY, predictions,
n_predictions, reg_type,
selected_y)
regression_errors_values(testY, predictions,
n_predictions)
save_train_test_data_xlsx(trainX, trainY, testX,
testY, predictions)
reg_type = "Случайный лес"
n_predictions = model_random_forest(
preprocessing.normalize(trainX), trainY,
preprocessing.normalize(testX), testY)
predictions = model_random_forest(
trainX, trainY, testX, testY)
regression_plot_show(trainX, testY, predictions,
n_predictions, reg_type,
selected_y)
regression_errors_values(testY, predictions,
n_predictions)
save_train_test_data_xlsx(trainX, trainY, testX,
testY, predictions)
reg_type = "Lasso"
n_predictions = model_lasso(
preprocessing.normalize(trainX), trainY,
preprocessing.normalize(testX), testY)
predictions = model_lasso(trainX, trainY, testX,
testY)
regression_plot_show(trainX, testY, predictions,
n_predictions, reg_type,
selected_y)
regression_errors_values(testY, predictions,
n_predictions)
save_train_test_data_xlsx(trainX, trainY, testX,
testY, predictions)
reg_type = ''
if reg_type == "Линейная регрессия":
n_predictions = model_linear_Regression(
preprocessing.normalize(trainX), trainY,
preprocessing.normalize(testX), testY)
predictions = model_linear_Regression(
trainX, trainY, testX, testY)
regression_plot_show(trainX, testY, predictions,
n_predictions, reg_type,
selected_y)
regression_errors_values(testY, predictions,
n_predictions)
save_train_test_data_xlsx(trainX, trainY, testX,
testY, predictions)
if reg_type == "Ridge":
n_predictions = model_ridge_Regression(
preprocessing.normalize(trainX), trainY,
preprocessing.normalize(testX), testY)
predictions = model_ridge_Regression(
trainX, trainY, testX, testY)
regression_plot_show(trainX, testY, predictions,
n_predictions, reg_type,
selected_y)
regression_errors_values(testY, predictions,
n_predictions)
save_train_test_data_xlsx(trainX, trainY, testX,
testY, predictions)
if reg_type == "Lasso":
n_predictions = model_lasso(
preprocessing.normalize(trainX), trainY,
preprocessing.normalize(testX), testY)
predictions = model_lasso(trainX, trainY, testX,
testY)
regression_plot_show(trainX, testY, predictions,
n_predictions, reg_type,
selected_y)
regression_errors_values(testY, predictions,
n_predictions)
save_train_test_data_xlsx(trainX, trainY, testX,
testY, predictions)
if reg_type == "Случайный лес":
n_predictions = model_random_forest(
preprocessing.normalize(trainX), trainY,
preprocessing.normalize(testX), testY)
predictions = model_random_forest(
trainX, trainY, testX, testY)
regression_plot_show(trainX, testY, predictions,
n_predictions, reg_type,
selected_y)
regression_errors_values(testY, predictions,
n_predictions)
save_train_test_data_xlsx(trainX, trainY, testX,
testY, predictions)
else:
st.warning("Выберите стоббец у")
elif problem_type == "Кластеризация":
n_clusters = st.number_input("Введите количество кластеров:",
3)
dataset = data.copy()
scaler = StandardScaler()
X = scaler.fit_transform(dataset)
km = KMeans(n_clusters=n_clusters)
# fit & predict clusters
dataset['cluster'] = km.fit_predict(X)
st.write(dataset['cluster'])
st.write(dataset)
if st.checkbox("PCA"):
pca = decomposition.PCA(n_components=2)
X_reduced = pca.fit_transform(X)
plt.figure(figsize=(6, 6))
plt.scatter(X_reduced[:, 0],
X_reduced[:, 1],
edgecolor='none',
alpha=0.7,
s=40,
c=dataset['cluster'],
cmap=plt.cm.get_cmap('nipy_spectral', 10))
plt.colorbar()
plt.title('Feautures PCA projection')
st.pyplot(plt.show())
if st.checkbox("TSNE"):
tsne = TSNE(random_state=17)
tsne_representation = tsne.fit_transform(X)
plt.figure(figsize=(6, 6))
plt.scatter(tsne_representation[:, 0],
tsne_representation[:, 1],
edgecolor='none',
alpha=0.7,
s=40,
c=dataset['cluster'],
cmap=plt.cm.get_cmap('nipy_spectral', 10))
plt.colorbar()
plt.title('Feautures T-sne projection ')
st.pyplot(plt.show())
save_exel(dataset, "Сохранить результаты xlsx File")
elif problem_type == "Классификация":
classification_type = st.radio(
"Выберите алгоритм решения",
("Все", "KNeighborsClassifier", "SVC_model"))
st.write("Выберите столбец у для задачи классификации")
selected_y = st.multiselect("", all_clummn_names)
if st.button("Выбрать"):
st.success("Столбец {} выбран успешно".format(selected_y))
# ".iloc" принимает row_indexer, column_indexer
y = np.array(data[selected_y])
X = np.array(data.drop(selected_y, 1))
# test_size показывает, какой объем данных нужно выделить для тестового набора
# Random_state — просто сид для случайной генерации
# Этот параметр можно использовать для воссоздания определённого результата:
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.20, random_state=27)
if classification_type == "Все":
SVC_model = svm.SVC()
st.write(
"В KNN-модели нужно указать параметр n_neighbors. Это число точек, на которое будет смотреть классификатор, чтобы определить, к какому классу принадлежит новая точка"
) #
nbr = st.slider("Число точек ", 3, data.shape[0])
KNN_model = KNeighborsClassifier(n_neighbors=nbr)
SVC_model.fit(X_train, y_train)
KNN_model.fit(X_train, y_train)
SVC_prediction = SVC_model.predict(X_test)
KNN_prediction = KNN_model.predict(X_test)
# Оценка точности — простейший вариант оценки работы классификатора
st.write("Оценка точности классификатора SVC_model")
st.write(accuracy_score(SVC_prediction, y_test))
st.write(
"Матрица неточности и отчёт о классификации дадут больше информации о производительности"
)
st.write(confusion_matrix(SVC_prediction, y_test))
st.write(classification_report(SVC_prediction, y_test))
st.write(SVC_prediction.tolist())
st.write(
"Оценка точности классификатора KNeighborsClassifier")
st.write(accuracy_score(KNN_prediction, y_test))
st.write(
"Матрица неточности и отчёт о классификации дадут больше информации о производительности"
)
# Но матрица неточности и отчёт о классификации дадут больше информации о производительности
st.write(confusion_matrix(KNN_prediction, y_test))
st.write(classification_report(KNN_prediction, y_test))
st.write(KNN_prediction.tolist())
if classification_type == "KNeighborsClassifier":
st.write(
"В KNN-модели нужно указать параметр n_neighbors. Это число точек, на которое будет смотреть классификатор, чтобы определить, к какому классу принадлежит новая точка"
) #
nbr = st.slider("Число точек ", 3, data.shape[0])
KNN_model = KNeighborsClassifier(n_neighbors=nbr)
KNN_model.fit(X_train, y_train)
KNN_prediction = KNN_model.predict(X_test)
st.write(
"Оценка точности классификатора KNeighborsClassifier")
st.write(accuracy_score(KNN_prediction, y_test))
st.write(
"Матрица неточности и отчёт о классификации дадут больше информации о производительности"
)
# Но матрица неточности и отчёт о классификации дадут больше информации о производительности
st.write(confusion_matrix(KNN_prediction, y_test))
st.write(classification_report(KNN_prediction, y_test))
st.write(KNN_prediction.tolist())
if classification_type == "SVC_model":
SVC_model = svm.SVC()
SVC_model.fit(X_train, y_train)
SVC_prediction = SVC_model.predict(X_test)
# Оценка точности — простейший вариант оценки работы классификатора
st.write("Оценка точности классификатора SVC_model")
st.write(accuracy_score(SVC_prediction, y_test))
st.write(
"Матрица неточности и отчёт о классификации дадут больше информации о производительности"
)
st.write(confusion_matrix(SVC_prediction, y_test))
st.write(classification_report(SVC_prediction, y_test))
st.write(SVC_prediction.tolist())
if st.button("Завершить работу"):
st.balloons()
def main():
st.write("""
# Heart Failure Prediction App
This app predicts the **Heart Failure** for a patient.
Data obtained from [here](https://bmcmedinformdecismak.biomedcentral.com/articles/10.1186/s12911-020-1023-5).
""")
activities = ["EDA", "Plot", "Model Building", "Predict", "About"]
choice = st.sidebar.selectbox("Select Activity", activities)
if choice == 'EDA':
st.subheader("Exploratory Data Analysis")
df = pd.read_csv('heart_failure_clinical_records_dataset.csv')
# data = st.file_uploader("Upload Dataset", type=["csv", "txt"])
# if data is not None:
# df = pd.read_csv(data)
st.dataframe(df)
if st.checkbox("Show shape"):
st.write(df.shape)
if st.checkbox("Show columns"):
all_columns = df.columns.to_list()
st.write(all_columns)
if st.checkbox("Select Columns To Show"):
selected_columns = st.multiselect("Select Columns", all_columns)
new_df = df[selected_columns]
st.dataframe(new_df)
if st.checkbox("Show summary"):
st.write(df.describe())
if st.checkbox("Show value counts"):
st.write(df.iloc[:, -1].value_counts())
elif choice == 'Plot':
st.subheader("Data Visualization")
df = pd.read_csv('heart_failure_clinical_records_dataset.csv')
st.dataframe(df)
if st.checkbox("Correlation with Seaborn"):
corr = df.corr()
st.write(sns.heatmap(corr))
st.pyplot()
if st.checkbox("Pie Chart"):
all_columns = df.columns.to_list()
colums_to_plot = st.selectbox("Select 1 column to plot",
all_columns)
pie_plot = df[colums_to_plot].value_counts().plot.pie()
st.write(pie_plot)
st.pyplot()
all_columns = df.columns.tolist()
type_of_plot = st.selectbox(
"Select Type of Plot",
["area", "bar", "line", "hist", "box", "kde"])
selected_columns_names = st.multiselect("selct Columns To plot",
all_columns)
if st.button("Generate Plot"):
st.success("Generating Customizable Plot of {} for {}".format(
type_of_plot, selected_columns_names))
# Plot by streamlit
if type_of_plot == "area":
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == "bar":
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == "line":
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
# Custom plot
elif type_of_plot:
cust_plot = df[selected_columns_names].plot(kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
elif choice == 'Model Building':
st.subheader("Building ML Model")
df = pd.read_csv('heart_failure_clinical_records_dataset.csv')
st.dataframe(df)
# Model building
X = df.iloc[:, 0:-1]
Y = df.iloc[:, -1]
seed = 7
# Model
models = []
# models.append(("LR", LogisticRegression()))
# models.append(("LDA", LinearDiscriminantAnalysis()))
# models.append(("KNN", KNeighborsClassifier()))
# models.append(('CART', DecisionTreeClassifier()))
# models.append(('NB', GaussianNB()))
# models.append(('SVM', SVC(probability=True)))
models.append(('RFC', RandomForestClassifier(n_estimators=100)))
# models.append(('GBC', GradientBoostingClassifier()))
# evaluate each model in turn
# List
model_name = []
model_mean = []
model_std = []
all_models = []
scoring = 'accuracy'
for name, model in models:
kfold = model_selection.KFold(n_splits=10, random_state=seed)
cv_results = model_selection.cross_val_score(model,
X,
Y,
cv=kfold,
scoring=scoring)
model_name.append(name)
model_mean.append(cv_results.mean())
model_std.append(cv_results.std())
accuracy_results = {
"model_name": name,
"model_accuracy": cv_results.mean(),
"standard_deviation": cv_results.std()
}
all_models.append(accuracy_results)
model.fit(X, Y)
pickle.dump(model, open(name + '.pkl', 'wb'))
st.dataframe(
pd.DataFrame(
zip(model_name, model_mean, model_std),
columns=['Model Name', 'Model Accuracy',
'Standard Deviation']))
st.json(all_models)
elif choice == 'Predict':
models = []
# models.append("LR.pkl")
# models.append("LDA.pkl")
# models.append("KNN.pkl")
# models.append('CART.pkl')
# models.append('NB.pkl')
# models.append('SVM.pkl')
models.append('RFC.pkl')
# models.append('GBC.pkl')
def user_input_features():
age = st.slider('Age of the patient(Years)', 40, 95, 50)
anaemia = st.selectbox(
'Anaemia-Decrease of red blood cells or hemoglobin(True-1, False-0)',
(1, 0))
creatinine_phosphokinase = st.slider(
'Creatinine phosphokinase-Level of the CPK enzyme in the blood(mcg/L)',
23, 7861, 3300)
diabetes = st.selectbox(
'Diabetes-If the patient has diabetes(True-1, False-0)',
(1, 0))
ejection_fraction = st.slider(
'Ejection fraction-Percentage of blood leaving', 14, 80, 30)
high_blood_pressure = st.selectbox(
'High blood pressure-If a patient has hypertension(True-1, False-0)',
(1, 0))
platelets = st.slider(
'Platelets-Platelets in the blood(kiloplatelets/mL)', 25100,
850000, 40000)
serum_creatinine = st.slider(
'Serum creatinine-Level of creatinine in the blood(mg/dL)',
0.5000, 9.4000, 1.2000)
serum_sodium = st.slider(
'Serum sodium-Level of sodium in the blood(mEq/L)', 113, 148,
120)
sex = st.selectbox('Sex-Woman or Man(Man-1,Women-0)', (1, 0))
smoking = st.selectbox(
'Smoking-If the patient smokes(True-1, False-0)', (1, 0))
time = st.slider('Time-Follow-up period(Days)', 4, 285, 100)
data = {
'age': age,
'anaemia': anaemia,
'creatinine_phosphokinase': creatinine_phosphokinase,
'diabetes': diabetes,
'ejection_fraction': ejection_fraction,
'high_blood_pressure': high_blood_pressure,
'platelets': platelets,
'serum_creatinine': serum_creatinine,
'serum_sodium': serum_sodium,
'sex': sex,
'smoking': smoking,
'time': time
}
features = pd.DataFrame(data, index=[0])
return features
input_df = user_input_features()
if input_df is not None:
st.dataframe(input_df)
for name in models:
model = pickle.load(open(name, 'rb'))
prediction_proba = model.predict_proba(input_df)
st.write('{} Predictions:'.format(name[0:-4]))
st.write(prediction_proba)
index = np.argmax(prediction_proba)
if index == 0:
st.write('Not Dead')
else:
st.write('Dead')
elif choice == 'About':
st.subheader("About")
st.write("Made By Rishab Koul with the Streamlit Library")
def main():
st.header("project Dataset explorer")
st.sidebar.header("OPTIONS")
all_cols = df.columns.values
numeric_cols = df.select_dtypes(include=numerics).columns.values
obj_cols = df.select_dtypes(include=["object"]).columns.values
if st.sidebar.checkbox("Data preview", True):
st.subheader("Data preview")
st.markdown(
f"Shape of dataset : {df.shape[0]} rows, {df.shape[1]} columns")
if st.checkbox("Data types"):
st.dataframe(df.dtypes)
if st.checkbox("Data Summary"):
st.write(df.describe())
if st.sidebar.checkbox("Pattern distribution", False):
st.subheader("Plot numeric column distribution")
with st.echo():
col = st.selectbox("Choose a column to display", numeric_cols)
n_bins = st.number_input("Max number of bins ?", 5, 100, 10)
chart = (alt.Chart(df).mark_bar().encode(
alt.X(f"{col}:Q", bin=alt.Bin(maxbins=n_bins)),
alt.Y("count()")))
st.altair_chart(chart)
st.markdown("---")
if st.sidebar.checkbox("Scatterplot", False):
st.subheader("Scatterplot")
selected_cols = st.multiselect("Choose 2 columns :", numeric_cols)
if len(selected_cols) == 2:
color_by = st.selectbox("Color by column:",
all_cols,
index=len(all_cols) - 1)
col1, col2 = selected_cols
chart = (alt.Chart(df).mark_circle(size=20).encode(
alt.X(f"{col1}:Q"), alt.Y(f"{col2}:Q"),
alt.Color(f"{color_by}")).interactive())
st.altair_chart(chart)
st.markdown("---")
# seaborn plot
if st.sidebar.checkbox("Correlation plot"):
st.subheader("Correlation plot")
cor = df.corr()
mask = np.zeros_like(cor)
mask[np.triu_indices_from(mask)] = True
plt.figure(figsize=(12, 10))
with sns.axes_style("white"):
st.write(
sns.heatmap(cor,
annot=True,
linewidth=2,
mask=mask,
cmap="magma"))
st.pyplot()
# Pie plot
if st.sidebar.checkbox("pie plot"):
st.subheader("Pie plot")
all_columns_names = df.columns.tolist()
st.success("Generating A pie plot")
st.write(df.iloc[:, -1].value_counts().plot.pie(autopct="%1.1f%%"))
st.pyplot()
if st.sidebar.checkbox("plot of value counts"):
st.subheader("Groupby columns")
st.text("value counts by target")
all_columns_name = df.columns.tolist()
primary_col = st.selectbox("primary columns to groupby",
all_columns_name)
selected_columns_name = st.multiselect("select columns to plot",
all_columns_name)
if st.button("plot"):
st.text("Generate value plot")
if selected_columns_name:
vc_plot = df.groupby(
primary_col)[selected_columns_name].count()
else:
vc_plot = df.iloc[:, -1].value_counts()
st.write(vc_plot.plot(kind="bar"))
st.pyplot()
# customizabe plot
if st.sidebar.checkbox("customizable plot", False):
st.subheader("Deviation")
columns_names = df.columns.tolist()
type_of_plot = st.selectbox(
"select type of plot",
["area", "bar", "line", "hist", "box", "kde"])
selected_columns_names = st.multiselect("select column to plot",
columns_names)
if st.button("Show plot"):
st.success("Generating customizable plot of {} for {}".format(
type_of_plot, selected_columns_names))
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
elif type_of_plot:
cust_plot = df[selected_columns_names].plot(
kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
if st.sidebar.checkbox("Deviations"):
st.subheader("Deviation plot")
for feature in ['time', 'measurement', 'control_mode']:
ax = plt.subplot()
st.write(
sns.distplot(df[feature][df.binary_result == 1],
bins=50,
label='Anormal',
kde_kws={'bw': 0.02}))
st.write(
sns.distplot(df[feature][df.binary_result == 0],
bins=50,
label='Normal',
kde_kws={'bw': 0.02}))
ax.set_xlabel('')
ax.set_title('histogram of feature: ' + str(feature))
plt.legend(loc='best')
st.pyplot()
def ztest(feature):
mean = falsepositive[feature].mean()
std = falsepositive[feature].std()
zScore = (falsenegative[feature].mean() -
mean) / (std / np.sqrt(sample_size))
return zScore
columns = df.drop('binary_result', axis=1).columns
falsepositive = df[df.binary_result == 0]
falsenegative = df[df.binary_result == 1]
sample_size = len(falsepositive)
significant_features = ["measurement"]
setpoint = 70
for i in columns:
z_value = ztest(i)
if (abs(z_value) >= setpoint):
st.write(i, " is critical alarm")
significant_features.append(i)
st.subheader("Inliers & Outliers of Data")
significant_features.append('binary_result')
y = df[significant_features]
inliers = df[df.binary_result == 0]
ins = inliers.drop(['binary_result'], axis=1)
outliers = df[df.binary_result == 1]
outs = outliers.drop(['binary_result'], axis=1)
ins.shape, outs.shape
def falsepositive_accuracy(values):
tp = list(values).count(1)
total = values.shape[0]
accuracy = np.round(tp / total, 4)
return accuracy
def falsenegative_accuracy(values):
tn = list(values).count(-1)
total = values.shape[0]
accuracy = np.round(tn / total, 4)
return accuracy
st.subheader("Accuracy score For Isolation forest")
ISF = IsolationForest(random_state=42)
ISF.fit(ins)
falsepositive_isf = ISF.predict(ins)
falsenegative_isf = ISF.predict(outs)
in_accuracy_isf = falsepositive_accuracy(falsepositive_isf)
out_accuracy_isf = falsenegative_accuracy(falsenegative_isf)
st.write("Accuracy in Detecting falsepositive Alarm:", in_accuracy_isf)
st.write("Accuracy in Detecting falsenegative Alarm:",
out_accuracy_isf)
st.subheader("Accuracy score For Local Outlier Factor")
LOF = LocalOutlierFactor(novelty=True)
LOF.fit(ins)
falsepositive_lof = LOF.predict(ins)
falsenegative_lof = LOF.predict(outs)
in_accuracy_lof = falsepositive_accuracy(falsepositive_lof)
out_accuracy_lof = falsenegative_accuracy(falsenegative_lof)
st.write("Accuracy in Detecting falsepositive Alarm :",
in_accuracy_lof)
st.write("Accuracy in Detecting falsenegative Alarm:",
out_accuracy_lof)
if st.sidebar.checkbox("Alarm Report", False):
st.subheader("classification of Alarm")
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=[16, 3])
ax1.set_title("Accuracy of Isolation Forest", fontsize=20)
st.write(
sns.barplot(x=[in_accuracy_isf, out_accuracy_isf],
y=['falsepositive Alarm', 'falsenegative Alarm'],
label="classifiers",
color="b",
ax=ax1))
ax1.set(xlim=(0, 1))
ax2.set_title("Accuracy of Local Outlier Factor", fontsize=20)
st.write(
sns.barplot(x=[in_accuracy_lof, out_accuracy_lof],
y=['falsepositive Alarm', 'falsenegative Alarm'],
label="classifiers",
color="r",
ax=ax2))
ax2.set(xlim=(0, 1))
st.pyplot()
rn = st.slider("", 1 , 15 , 5)
df , name = GetPerformanceStat(int(PL_ID), rn)
st.header(name + "'s Performance Stats")
st.subheader('Select the options you want to see stats for')
chioce = st.radio('', ('Overall','Raids','Tackles') )
if chioce == 'Overall':
# Total points stats
st.subheader('Total points stats in recent '+str(rn)+' Matches')
if st.checkbox('Show Data'):
st.write(df[['player_total_points','player_raid_points_total','player_tackle_points_total']].T )
if st.checkbox('Show Bar Chart'):
st.bar_chart(df[['player_total_points','player_raid_points_total','player_tackle_points_total']] )
if st.checkbox('Show Area Chart'):
st.area_chart(df[['player_total_points','player_raid_points_total','player_tackle_points_total']])
elif chioce == 'Raids':
# Raids Stats
st.subheader('Total Raids Statsin recent '+str(rn)+' Matches')
if st.checkbox('Show Data'):
st.write(df[['player_raids_total','player_raids_successful','player_raids_empty','player_raids_unsuccessful']].T )
if st.checkbox('Show Bar Chart'):
st.bar_chart(df[['player_raids_total','player_raids_successful','player_raids_empty','player_raids_unsuccessful']] )
if st.checkbox('Show Area Chart'):
st.area_chart(df[['player_raids_total','player_raids_successful','player_raids_empty','player_raids_unsuccessful']] )
elif chioce == 'Tackles':
# Tackel stats
st.subheader('Total Raids Statsin recent '+str(rn)+' Matches')
if st.checkbox('Show Data'):
def main():
""" BPX Energy """
st.title("BPX Energy LE Forecast Tool")
st.subheader("SoHa Change Dev/Test")
html_temp = """
<div style="background-color:green;"><p style="color:white;font-size:50px;padding:10px">SoHa Change is Awesome</p></div>
"""
st.markdown(html_temp,unsafe_allow_html=True)
def file_selector(folder_path='./datasets'):
filenames = os.listdir(folder_path)
selected_filename = st.selectbox("Select A file",filenames)
return os.path.join(folder_path,selected_filename)
filename = file_selector()
st.info("You Selected {}".format(filename))
# Read Data
df = pd.read_csv(filename)
# Show Dataset
if st.checkbox("Show Dataset"):
number = st.number_input("Number of Rows to View")
st.dataframe(df.head(n=9))
# Show Columns
if st.button("Column Names"):
st.write(df.columns)
# Show Shape
if st.checkbox("Shape of Dataset"):
data_dim = st.radio("Show Dimension By ",("Rows","Columns"))
if data_dim == 'Rows':
st.text("Number of Rows")
st.write(df.shape[0])
elif data_dim == 'Columns':
st.text("Number of Columns")
st.write(df.shape[1])
else:
st.write(df.shape)
# Select Columns
if st.checkbox("Select Columns To Show"):
all_columns = df.columns.tolist()
selected_columns = st.multiselect("Select",all_columns)
new_df = df[selected_columns]
st.dataframe(new_df)
# Show Values
if st.button("Value Counts"):
st.text("Value Counts By Target/Class")
st.write(df.iloc[:,-1].value_counts())
# Show Datatypes
if st.button("Data Types"):
st.write(df.dtypes)
# Show Summary
if st.checkbox("Summary"):
st.write(df.describe().T)
## Plot and Visualization
st.subheader("Data Visualization")
# Correlation
# Seaborn Plot
if st.checkbox("Correlation Plot[Seaborn]"):
st.write(sns.heatmap(df.corr(),annot=True))
st.pyplot()
# Pie Chart
if st.checkbox("Pie Plot"):
all_columns_names = df.columns.tolist()
if st.button("Generate Pie Plot"):
st.success("Generating A Pie Plot")
st.write(df.iloc[:,-1].value_counts().plot.pie(autopct="%1.1f%%"))
st.pyplot()
# Count Plot
if st.checkbox("Plot of Value Counts"):
st.text("Value Counts By Target")
all_columns_names = df.columns.tolist()
primary_col = st.selectbox("Primary Columm to GroupBy",all_columns_names)
selected_columns_names = st.multiselect("Select Columns",all_columns_names)
if st.button("Plot"):
st.text("Generate Plot")
if selected_columns_names:
vc_plot = df.groupby(primary_col)[selected_columns_names].count()
else:
vc_plot = df.iloc[:,-1].value_counts()
st.write(vc_plot.plot(kind="bar"))
st.pyplot()
# Customizable Plot
st.subheader("Customizable Plot")
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox("Select Type of Plot",["area","bar","line","hist","box","kde"])
selected_columns_names = st.multiselect("Select Columns To Plot",all_columns_names)
if st.button("Generate Plot"):
st.success("Generating Customizable Plot of {} for {}".format(type_of_plot,selected_columns_names))
# Plot By Streamlit
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
# Custom Plot
elif type_of_plot:
cust_plot= df[selected_columns_names].plot(kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
if st.button("SoHa Change"):
st.balloons()
st.sidebar.header("LE Utility Tool")
st.sidebar.info("The LE & Forecast Utility is a tool meant to improve the speed & consistency of production reporting for operations PEs.")
st.sidebar.header("Get Datasets")
st.sidebar.markdown("[Common ML Dataset Repo]("")")
st.sidebar.header("Developed By")
st.sidebar.info("*****@*****.**")
st.sidebar.info("*****@*****.**")
st.sidebar.text("SoHa Change Team lead: Kellen McLoughlin")
st.sidebar.text("Web App maintained by MarkusJBPX")
""")
st.write("""
## Giá mở
""")
st.line_chart(df.Open)
st.write("""
## Giá đóng
""")
st.line_chart(df.Close)
st.write("""
## Số lượng giao dịch trong ngày
""")
st.area_chart(df.Volume)
file = st.file_uploader("Pick a file")
if file:
df1 = pd.read_csv(file)
# Lấy cột Volume
Test_1 = np.array(df1[['Volume']])
# Lấy cột Open
Test_2 = np.array(df1[['Open']])
# Lấy cột High
Test_3 = np.array(df1[['High']])
# Lấy cột Low
def main():
activities = ["EDA &VIZ", "Modelling"]
choice = st.sidebar.selectbox("Select Activities", activities)
if st.sidebar.checkbox('About'):
st.sidebar.markdown("""
app work in progress .This is a beta release.
version: b-0.0.1
initial release:27/6/2020
helpful suggestions are welcome.
contact: [email protected]
""")
if choice == 'EDA &VIZ':
st.title('Play with ML')
html_temp1 = """<img src="images/dobby1.jpeg" alt="It's dobby" width="120" height="150">"""
st.markdown(html_temp1, unsafe_allow_html=True)
html_temp = """
<div style="background-color:coral;padding:12px">
<h2 style="color:white;text-align:center;"> Play with ML App </h2>
</div>
"""
st.markdown(html_temp, unsafe_allow_html=True)
st.markdown(
'hey,tired of modelling and tuning ML Models, wanna play with data & ML modles? Then upload a dataset here.. **_Dobby , a free elf_** is here for you '
)
st.subheader("Exploratory Data Analysis & Vizualization ")
data = st.file_uploader("Upload a Dataset", type=["csv"])
if data is not None:
st.subheader('EDA')
df = pd.read_csv(data)
st.write('shape:', df.shape)
if st.checkbox("Show Columns"):
all_columns = df.columns.to_list()
#st.write(all_columns)
if st.checkbox("Null values"):
st.write(df.isnull().sum())
if st.checkbox("Information"):
st.write(df.info())
if st.checkbox("Summary"):
st.write(df.describe())
if st.checkbox("Show Selected Columns"):
all_columns_names = df.columns.tolist()
selected_columns = st.multiselect("Select Columns",
all_columns)
df1 = df[selected_columns]
#st.dataframe(df1)
if st.checkbox("Correlation Plot(Seaborn)"):
st.write(sns.heatmap(df.corr(), annot=True))
st.pyplot()
st.subheader('Data Visualization')
if st.checkbox("Show Value Counts"):
column = st.selectbox("Select a Column to show value counts",
all_columns)
st.write(df[column].value_counts())
st.write(df[column].value_counts().plot(kind='bar'))
st.pyplot()
all_columns_names = df.columns.tolist()
type_of_plot = st.selectbox("Select Type of Plot", [
"area", "bar", "pie", "line", "hist", "box", "kde",
"altair_chart"
])
selected_columns_names = st.multiselect("Select Columns To Plot",
all_columns_names)
if st.button("Generate Plot"):
st.success("Generating {} plot for {}".format(
type_of_plot, selected_columns_names))
if type_of_plot == 'area':
cust_data = df[selected_columns_names]
st.area_chart(cust_data)
st.pyplot()
elif type_of_plot == 'bar':
cust_data = df[selected_columns_names]
st.bar_chart(cust_data)
st.pyplot()
elif type_of_plot == "Pie Plot":
column_to_plot = st.selectbox("Select 1 Column",
selected_columns_names)
pie_plot = df[column_to_plot].value_counts().plot.pie(
autopct="%1.1f%%")
st.write(pie_plot)
st.pyplot()
elif type_of_plot == 'line':
cust_data = df[selected_columns_names]
st.line_chart(cust_data)
st.pyplot()
elif type_of_plot == 'altair_chart':
a = st.selectbox("Select X axis", all_columns)
b = st.selectbox("Select Y axis", all_columns)
c = st.selectbox("Select a column ", all_columns)
cust_data = pd.DataFrame([a, b, c])
c = alt.Chart(cust_data).mark_circle().encode(
x='a',
y='b',
size='c',
color='c',
tooltip=['a', 'b', 'c'])
st.altair_chart(c, use_container_width=True)
st.pyplot()
elif type_of_plot:
cust_plot = df[selected_columns_names].plot(
kind=type_of_plot)
st.write(cust_plot)
st.pyplot()
if choice == 'Modelling':
html_temp = """
<div style="background-color:coral;padding:12px">
<h2 style="color:white;text-align:center;"> Play with ML App </h2>
</div>
"""
st.header('Training')
st.markdown(
"**_Hello Iam Dobby. Dobby has no master - Dobby is a free elf_**. Due to SARS-CoV-2 lockdown I dont have much work to do , So Iam here to make your model."
)
data = st.file_uploader("Upload a Dataset", type=["csv"])
if data is not None:
st.subheader('EDA')
df = pd.read_csv(data)
st.dataframe(df.head())
st.write('shape:', df.shape)
st.header('Data Preprocessing')
all_columns = df.columns.tolist()
features = st.multiselect("Select feature columns", all_columns)
X = df[features]
st.dataframe(X)
st.write(X.head())
st.write(X.shape)
labels = st.selectbox("Select label column", all_columns)
y = df[labels]
st.dataframe(y)
st.write(y.head())
st.write(X.shape)
all_columns = X.columns.tolist()
if st.checkbox("Handling missing values"):
radioval = st.radio("choose type", ('ffill', 'statistical'))
if radioval == 'None':
print('handling missing values skipped')
elif radioval == 'fbfill':
if st.checkbox("fbfill"):
X = X.ffill(axis=0)
X = X.ffill(axis=0)
st.markdown('**_missing values are fb filled_**')
elif radioval == 'statistical':
if st.checkbox("handle with mean"):
selected_columns = st.multiselect(
"Select Columns to handle with mean ", all_columns)
X[selected_columns] = X[selected_columns].fillna(
X[selected_columns].mean(), inplace=True)
st.write('handled with mean')
elif st.checkbox("handle with median"):
selected_columns = st.multiselect(
"Select Columns to handle with median",
all_columns)
X[selected_columns] = X[selected_columns].fillna(
X[selected_columns].median(), inplace=True)
st.write('handled with median')
elif st.checkbox("handle with mode"):
selected_columns = st.multiselect(
"Select Columns to handle with mode", all_columns)
X[selected_columns] = X[selected_columns].fillna(
X[selected_columns].mode()[0], inplace=True)
st.write('handled with mode')
st.markdown(
'**_missing values are filled statistically_**')
st.write('missing values:', X.isnull().sum())
if st.checkbox("One hot encoding"):
if st.checkbox("encode features"):
X = pd.get_dummies(X)
st.write("features are one hot encoded")
if st.checkbox("encode labels"):
y = pd.get_dummies(y)
st.write("labels are one hot encoded")
st.dataframe(y)
st.write('Train - val split')
number = st.number_input('test split size',
min_value=0.1,
max_value=1.00)
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=number, random_state=0)
st.write(X_train.shape)
st.write(X_test.shape)
if st.checkbox("Feature Scaling"):
radioval = st.radio(
"choose type of feature scaling",
('none', 'Standardization', 'Normalization'))
if radioval == 'none':
st.write("you skipped feature scaling")
if radioval == 'Standardization':
from sklearn.preprocessing import StandardScaler
sc_X = StandardScaler()
X_train = sc_X.fit_transform(X_train)
X_test = sc_X.transform(X_test)
#sc_y = StandardScaler()
#y_train = sc_y.fit_transform(y_train)
if radioval == 'Normalization':
min_max_scaler = sklearn.preprocessing.MinMaxScaler()
X_train = min_max_scaler.fit_transform(X_train)
X_test = min_max_scaler.transform(X_test)
st.header("Training")
problem_types = ['Regression', 'Classification']
problem_type = st.selectbox("Select Problem Type ", problem_types)
st.sidebar.markdown("Hyperparameter Tuning")
if problem_type == 'Classification':
models = [
'Logistic Regression', 'KNN', 'SVM', 'DecisionTree',
'Random Forest', 'XgBoostClassifier'
]
model = st.selectbox("Select a model ", models)
if model == 'Logistic Regression':
from sklearn.linear_model import LogisticRegression
classifier = LogisticRegression(random_state=0)
if model == 'KNN':
n_neighbors = st.sidebar.slider('n_neighbors',
min_value=1,
max_value=5,
step=1)
p = st.sidebar.selectbox("P", [1, 2, 3, 4])
from sklearn.neighbors import KNeighborsClassifier
classifier = KNeighborsClassifier(n_neighbors=n_neighbors,
metric='minkowski',
p=p)
if model == 'SVM':
from sklearn.svm import SVC
kernel_list = ['linear', 'poly', 'rbf', 'sigmoid']
kernel = st.sidebar.selectbox("P", kernel_list)
C = st.sidebar.slider('C',
min_value=1,
max_value=6,
step=1)
degree = st.sidebar.slider('Degree',
min_value=1,
max_value=10,
step=1)
classifier = SVC(kernel=kernel,
C=C,
random_state=0,
degree=degree)
if model == 'DecisionTree':
from sklearn.tree import DecisionTreeClassifier
criterion = st.sidebar.selectbox("criterion",
["gini", "entropy"])
max_depth = st.sidebar.slider('max_depth',
min_value=1,
max_value=10,
step=1)
min_samples_leaf = st.sidebar.slider('min_samples_leaf',
min_value=1,
max_value=10,
step=1)
classifier = DecisionTreeClassifier(
criterion=criterion,
max_depth=max_depth,
min_samples_leaf=min_samples_leaf,
random_state=0)
if model == 'Random Forest':
from sklearn.ensemble import RandomForestClassifier
criterion = st.sidebar.selectbox("criterion",
["gini", "entropy"])
n_estimators = st.sidebar.number_input('n_estimators',
min_value=1,
max_value=500,
step=1)
max_depth = st.sidebar.slider('max_depth',
min_value=1,
max_value=10,
step=1)
classifier = RandomForestClassifier(
n_estimators=n_estimators,
criterion=criterion,
max_depth=max_depth,
random_state=0)
if model == 'XgBoostClassifier':
from xgboost import XGBClassifier
n_estimators = st.sidebar.number_input('n_estimators',
min_value=1,
max_value=2000)
reg_lambda = st.sidebar.number_input('reg_lambda',
min_value=0.01,
max_value=5.00,
step=0.02)
max_depth = st.sidebar.slider('max_depth',
min_value=1,
max_value=10,
step=1)
colsample_bytree = st.sidebar.number_input(
'colsample_bytree',
min_value=0.50,
max_value=1.00,
step=0.05)
classifier = XGBClassifier(
n_estimators=n_estimators,
reg_lambda=reg_lambda,
max_depth=max_depth,
colsample_bytree=colsample_bytree)
if st.button("Train"):
with st.spinner('model is training...'):
classifier.fit(X_train, y_train)
st.success('Model trained!')
y_pred = classifier.predict(X_test)
from sklearn.metrics import accuracy_score
acc = accuracy_score(y_test, y_pred)
st.write('val_accuracy:', acc)
from sklearn.metrics import confusion_matrix, classification_report
st.write(classification_report(y_test, y_pred))
cm = confusion_matrix(y_test, y_pred)
st.markdown("**_confusion matrix_**")
st.write(cm)
y_pred = pd.DataFrame(y_pred)
st.dataframe(y_pred)
st.write(y_pred[0].value_counts())
st.write(y_pred[0].value_counts().plot(kind='bar'))
st.pyplot()
st.balloons()
def download_model(model):
output_model = pickle.dumps(model)
st.write("model saved as output_model ")
b64 = base64.b64encode(output_model).decode()
href = f'<a href="data:file/output_model;base64,{b64}">Download Trained Model</a>'
st.markdown(href, unsafe_allow_html=True)
if st.button("save & Download model"):
download_model(classifier)
if problem_type == 'Regression':
models = [
'Linear Regression', 'SVR', 'DecisionTree',
'Random Forest', 'XgBoostRegression'
]
model = st.selectbox("Select a model ", models)
if model == 'Linear Regression':
from sklearn.linear_model import LinearRegression
regressor = LinearRegression()
if model == 'SVR':
from sklearn.svm import SVR
kernel_list = ['linear', 'poly', 'rbf', 'sigmoid']
kernel = st.sidebar.selectbox("P", kernel_list)
degree = st.sidebar.slider('Degree',
min_value=1,
max_value=10,
step=1)
regressor = SVR(kernel=kernel, degree=degree)
if model == 'DecisionTree':
from sklearn.tree import DecisionTreeRegressor
criterion = st.sidebar.selectbox(
"criterion", ["mse", "friedman_mse", "mae"])
max_depth = st.sidebar.slider('max_depth',
min_value=1,
max_value=10,
step=1)
min_samples_leaf = st.sidebar.slider('min_samples_leaf',
min_value=1,
max_value=10,
step=1)
regressor = DecisionTreeRegressor(
criterion=criterion,
max_depth=max_depth,
min_samples_leaf=min_samples_leaf,
random_state=0)
if model == 'Random Forest':
from sklearn.ensemble import RandomForestRegressor
n_estimators = st.sidebar.number_input('n_estimators',
min_value=1,
max_value=500,
step=1)
max_depth = st.sidebar.slider('max_depth',
min_value=1,
max_value=10,
step=1)
criterion = st.sidebar.selectbox("criterion",
["mse", "mae"])
regressor = RandomForestRegressor(
n_estimators=n_estimators,
criterion=criterion,
max_depth=max_depth,
random_state=0)
if model == 'XgBoostRegression':
from xgboost import XGBRegressor
n_estimators = st.sidebar.number_input('n_estimators',
min_value=1,
max_value=2000)
reg_lambda = st.sidebar.number_input('reg_lambda',
min_value=0.01,
max_value=5.00,
step=0.02)
max_depth = st.sidebar.slider('max_depth',
min_value=1,
max_value=10,
step=1)
booster = st.sidebar.selectbox(
'booster', ["gbtree", "gblinear", "dart"])
learning_rate = st.sidebar.number_input('learning_rate',
min_value=0.05,
max_value=3.00,
step=0.01)
colsample_bytree = st.sidebar.number_input(
'colsample_bytree',
min_value=0.50,
max_value=1.00,
step=0.05)
regressor = XGBRegressor(n_estimators=n_estimators,
learning_rate=learning_rate,
booster=booster,
reg_lambda=reg_lambda,
max_depth=max_depth,
colsample_bytree=colsample_bytree)
if st.button("Train"):
with st.spinner('model is training...'):
regressor.fit(X_train, y_train)
st.success('Model trained!')
y_pred = regressor.predict(X_test)
from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score
mae_tr = mean_absolute_error(y_train,
regressor.predict(X_train))
mae = mean_absolute_error(y_test, y_pred)
mse = mean_squared_error(y_test, y_pred)
mse_tr = mean_squared_error(y_train,
regressor.predict(X_train))
r2 = r2_score(y_test, y_pred)
r2_tr = r2_score(y_train, regressor.predict(X_train))
st.write('mean absolute error:')
st.write('train:', mae_tr, 'val:', mae)
st.write('mean squared error:')
st.write('train:', mse_tr, 'val:', mse)
st.write('r2:')
st.write('train:', r2_tr, 'val:', r2)
y_pred = pd.DataFrame(y_pred)
st.dataframe(y_pred)
st.balloons()
def download_model(model):
output_model = pickle.dumps(model)
st.write("model saved as output_model ")
b64 = base64.b64encode(output_model).decode()
href = f'<a href="data:file/output_model;base64,{b64}">Download Trained Model</a>'
st.markdown(href, unsafe_allow_html=True)
if st.button("save & Download model"):
download_model(regressor)
df = pd.DataFrame({
'pool selection': ['Pool 1', 'Pool 2', 'Pool 3', 'Pool 4'],
})
option = st.sidebar.selectbox('Which pool do you want ?', df['pool selection'])
'You selected : ', option
latest_iteration = st.empty()
bar = st.progress(0)
array = [0.0]
if st.button('Compute'):
chart = st.area_chart(array)
for i in range(100):
# Random data to display
latest_iteration.text(f'Computing {i+1}%')
bar.progress(i + 1)
df = pd.DataFrame({
abs((np.random.randn() + 10) / 10),
})
chart.add_rows(df)
time.sleep(0.1)
st.write("Finished !")