#We are interested in looking at the confusion matrices
import header as h
import pandas as pd
import numpy as np
import fs_definitions as fsd
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.metrics import plot_confusion_matrix
from sklearn.metrics import confusion_matrix
df = h.create_master_df() #Create the master df
for fs in h.FEATURES:
temp_df = fsd.create_prepped_df(fs, df) #create the prepped df
y = temp_df['label'] #Set 'y' to the winner value
#Remove the winner and label from the features
temp_prepped_df = temp_df.drop(['Winner', 'label'], axis=1)
X = temp_prepped_df.values #Set x to the values
#class_names = temp_df.Winner #This should match to the winners?
class_names = ['Blue', 'Red']
print(class_names)
print(X.shape)
print(y.shape)
pd.set_option('display.max_rows', 500) #Used for debugging
#df = h.create_fight_df(h.MASTER_CSV_FILE)
#print(df.head)
#print(df.describe())
#print(len(df))
#print(df.head)
#print(df.dtypes)
#OK... Now we need to create some dataframes....
df = h.create_master_df()
#print(df.head)
#print(len(df))
#adding a comment
temp_df = fsd.create_prepped_df('c1', df)
#print(temp_df)
X = temp_df.iloc[:, :-2].values
y = temp_df.iloc[:, -1:].values
#print(X)
#print(y)
print(X.shape)
print(y.shape)
X_train, X_test, y_train, y_test = train_test_split(X,
y,
def get_recent_probs(df: pd.DataFrame, fs: str) -> pd.DataFrame:
"""
Returns a probability dataframe of the last 50 events.
Parameters
----------
df : pd.DataFrame
Master DataFrame
fs : str
The feature set
Returns
-------
pd.DataFrame
The probability DataFrame
"""
#We want to figure out how to split on the date....
prepped_df = fsd.create_prepped_df(fs, df)
list_of_dates = (prepped_df['date_final'].unique())
list_of_dates = (np.flip(np.sort(list_of_dates)))
list_of_dates = list_of_dates[:50]
print(list_of_dates)
y = prepped_df[['label', 'date_final']]
ev_df = prepped_df[[
'date_final', 'B_ev_final', 'R_ev_final', 'Winner', 'label',
'country_final'
]]
prepped_df = prepped_df.drop(
['Winner', 'label', 'R_ev_final', 'B_ev_final', 'country_final'],
axis=1)
final_probs = None
for d in list_of_dates:
X_test = prepped_df.loc[prepped_df['date_final'] == d]
X_train = prepped_df.loc[prepped_df['date_final'] != d]
y_test = y.loc[y['date_final'] == d]
y_train = y.loc[y['date_final'] != d]
X_test_ev = ev_df.loc[ev_df['date_final'] == d]
X_train_ev = ev_df.loc[ev_df['date_final'] != d]
#Remove the date
X_test = X_test.drop('date_final', 1)
X_train = X_train.drop('date_final', 1)
y_train = y_train.drop('date_final', 1)
y_test = y_test.drop('date_final', 1)
print(X_test.shape)
print(X_train.shape)
print(y_test.shape)
print(y_train.shape)
print(X_test_ev.shape)
print(X_train_ev.shape)
print()
print()
classifier = get_classifier(fs)
classifier.fit(X_train, y_train.values.ravel())
probs = classifier.predict_proba(X_test)
preds = classifier.predict(X_test)
preds = preds.reshape((len(preds), 1))
X_test = np.append(X_test, probs, 1)
X_test = np.append(X_test, preds, 1)
X_test = np.append(X_test, X_test_ev, 1)
trash_df = prepped_df.drop(['date_final'], axis=1)
colNamesArr = trash_df.columns.values
colNamesArr = np.append(colNamesArr, [
'B_prob', 'R_prob', 'preds', 'date_final', 'B_ev_final',
'R_ev_final', 'Winner', 'label', 'country_final'
])
final_df = pd.DataFrame(X_test)
final_df.columns = colNamesArr
if (final_probs is not None):
final_probs = final_probs.append(final_df)
else:
final_probs = final_df
return (final_probs)
pass
import pandas as pd
import numpy as np
import fs_definitions as fsd
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.linear_model import LogisticRegression
import seaborn as sns
import matplotlib.pyplot as plt
sns.set(style="white")
sns.set(style="whitegrid", color_codes=True)
np.set_printoptions(suppress=True)
df = h.create_fight_df('../data/ufc-master.csv')
fs = 'c1'
temp_df = fsd.create_prepped_df(fs, df)
print(df.shape)
print(temp_df.shape)
y = temp_df['label']
#WE NEED TO STRIP OUT R_ev_final and B_ev_final
#I THINK WE CAN DO THIS AND THEN SPLIT....
ev_df = temp_df[['B_ev_final', 'R_ev_final']]
temp_prepped_df = temp_df.drop(['Winner', 'label', 'R_ev_final', 'B_ev_final'],
axis=1)
X = temp_prepped_df.values
X_ev = ev_df.values
def get_test_probs(df: pd.DataFrame,
fs: str,
seed: int,
split: float,
count_split=False) -> pd.DataFrame:
"""
Returns a DataFrame that includes classification results
Parameters
----------
df : pd.DataFrame
The DataFrame we will be analyzing (Probably the master dataframe)
fs: feature set
The feature set
seed : int
A seed for the split
split : float
the split amount
Returns
-------
Returns a dataframe that has been put through a classifier function.
The dataframe will include the probabilities and predictions.
"""
print("\n\n******************************************************\n\n")
print("STARTING get_test_probs")
print("\n\n******************************************************")
"""
1. Create Prepped DF
2. Create alternate DF that only includes odds
3. Remove certain characteristics from Prepped DF
4. Split both DFs
4.5: Create classifier
5. Run classification
6. Append odds and probs and preds and winner and label to df
7. Return it.
"""
#1. create prepped df
prepped_df = fsd.create_prepped_df(fs, df)
#2. Create alternate DF that only includes odds
ev_df = prepped_df[[
'date_final', 'B_ev_final', 'R_ev_final', 'Winner', 'label',
'country_final'
]]
#3. Remove certain characteristics from Prepped DF
y = prepped_df['label']
prepped_df = prepped_df.drop([
'Winner', 'label', 'R_ev_final', 'B_ev_final', 'date_final',
'country_final'
],
axis=1)
X = prepped_df.values
X_ev = ev_df.values
#4. Split both DFs
if count_split == False:
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=split,
random_state=seed)
X_train_ev, X_test_ev, y_train_ev, y_test_ev = train_test_split(
X_ev, y, test_size=split, random_state=seed)
else:
X_test = X[:split]
X_train = X[split:]
y_test = y[:split]
y_train = y[split:]
X_test_ev = X_ev[:split]
X_train_ev = X_ev[split:]
y_test_ev = y[:split]
y_train_ev = y[split:]
#4.5 Create classifier
classifier = get_classifier(fs)
#5. Run classification
print(f"x-train shape is {X_train.shape}")
classifier.fit(X_train, y_train)
probs = classifier.predict_proba(X_test)
preds = classifier.predict(X_test)
#p_and_p = np.append(probs, preds, 1)
#6. Append odds and probs and preds and winner and label to df
preds = preds.reshape((len(preds), 1))
X_test = np.append(X_test, probs, 1)
X_test = np.append(X_test, preds, 1)
X_test = np.append(X_test, X_test_ev, 1)
colNamesArr = prepped_df.columns.values
colNamesArr = np.append(colNamesArr, [
'B_prob', 'R_prob', 'preds', 'date_final', 'B_ev_final', 'R_ev_final',
'Winner', 'label', 'country_final'
])
print(X_test.shape)
print(colNamesArr.shape)
final_df = pd.DataFrame(X_test)
final_df.columns = colNamesArr
return final_df