"""
# In[]
import HappyML.preprocessor as pp
dataset = pp.dataset("C:/Users/henry/Desktop/Python Training/Python機器學習/範例原始碼＆「快樂版」函式庫/Ch05 Regression/50_Startups.csv")
X, Y = pp.decomposition(dataset, [0, 1, 2, 3], [4])
# X = pp.onehot_encoder(X, columns=[3])
# X = pp.remove_columns(X, [3])
X = pp.onehot_encoder(X, columns=[3], remove_trap=True)
X_train, X_test, Y_train, Y_test = pp.split_train_test(X, Y, train_size=0.8)
X_train, X_test= pp.feature_scaling(fit_ary=X_train, transform_arys=(X_train, X_test))
Y_train, Y_test= pp.feature_scaling(fit_ary=Y_train, transform_arys=(Y_train, Y_test))
# In[]
from HappyML.regression import SimpleRegressor
simple_reg = SimpleRegressor()
Y_pred_simple = simple_reg.fit(X_train, Y_train).predict(X_test)
# R-Squared always increase in multiple linear regression --> Use Adjusted R-Squared instead
print("Goodness of Model (R-Squared Score):", simple_reg.r_score(X_test, Y_test))
# In[]
from HappyML.regression import MultipleRegressor
X_train = pp.add_constant(X_train)
""" # In[] Preprocessing import HappyML.preprocessor as pp # Load Dataset dataset = pp.dataset(file="Mall_Customers.csv") # Decomposition X = pp.decomposition(dataset, x_columns=[1, 2, 3, 4]) # One-Hot Encoding X = pp.onehot_encoder(ary=X, columns=[0], remove_trap=True) # Feature Scaling (for PCA Feature Selection) X = pp.feature_scaling(fit_ary=X, transform_arys=X) # Feature Selection (PCA) from HappyML.preprocessor import PCASelector selector = PCASelector() X = selector.fit(x_ary=X, verbose=True, plot=True).transform(x_ary=X) # In[] K-Means Clustering with Fixed Clusters = 4 (Without HappyML) # from sklearn.cluster import KMeans # import time # # K-Means Clustering with K=4 # kmeans = KMeans(n_clusters=4, init="k-means++", random_state=int(time.time())) # Y_pred = kmeans.fit_predict(X)
import HappyML.preprocessor as pp
from HappyML.classification import SVM
from HappyML.performance import KFoldClassificationPerformance
import numpy as np
from HappyML.performance import GridSearch
# SVM without GridSearch
dataset = pp.dataset("Voice.csv")
X, Y = pp.decomposition(dataset, [i for i in range(20)], [20])
Y, Y_mapping = pp.label_encoder(Y, mapping=True)
selector = pp.KBestSelector()
X = selector.fit(X, Y, True, True).transform(X)
X_train, X_test, Y_train, Y_test = pp.split_train_test(X, Y)
X_train, X_test = pp.feature_scaling(X_train, (X_train, X_test))
classifier = SVM()
Y_pred = classifier.fit(X_train, Y_train).predict(X_test)
K = 10
kfp = KFoldClassificationPerformance(X, Y, classifier.classifier, K)
print("----- SVM Classification -----")
print("{} Folds Mean Accuracy: {}".format(K, kfp.accuracy()))
print("{} Folds Mean Recall: {}".format(K, kfp.recall()))
print("{} Folds Mean Precision: {}".format(K, kfp.precision()))
print("{} Folds Mean F1-score: {}".format(K, kfp.f_score()))
# SVM with GridSearch
X, Y = pp.decomposition(dataset, [i for i in range(20)], [20])
"""
import HappyML.preprocessor as pp
from HappyML.clustering import KMeansCluster
import HappyML.model_drawer as md
from HappyML.classification import DecisionTree
from HappyML.performance import KFoldClassificationPerformance
from IPython.display import Image, display
dataset = pp.dataset("CreditCards.csv")
dataset = pp.missing_data(dataset)
X = pp.decomposition(dataset, [i for i in range(18) if i != 0])
X = pp.feature_scaling(X, X)
selector = pp.PCASelector(best_k=2)
X = selector.fit(X).transform(X)
cluster = KMeansCluster()
Y_pred = cluster.fit(X).predict(X, "Customer Type")
md.cluster_drawer(X, Y_pred, cluster.centroids, "Customers Segmentation",
"Microsoft JhengHei")
dataset = pp.combine(dataset, Y_pred)
X, Y = pp.decomposition(dataset, [i for i in range(18) if i != 0], [18])
selector = pp.KBestSelector()
# In[] Import & Load data
import HappyML.preprocessor as pp
dataset = pp.dataset(file="CarEvaluation.csv")
# In[] Decomposition
X, Y = pp.decomposition(dataset,
x_columns=[i for i in range(4)],
y_columns=[4])
# In[] Missing Data
X = pp.missing_data(X, strategy="mean")
# In[] Categorical Data Encoding
# Label Encoding
Y, Y_mapping = pp.label_encoder(Y, mapping=True)
# One-Hot Encoding
X = pp.onehot_encoder(X, columns=[0])
# In[] Split Training Set, Testing Set
X_train, X_test, Y_train, Y_test = pp.split_train_test(X,
Y,
train_size=0.8,
random_state=0)
# In[] Feature Scaling for X_train, X_test
X_train, X_test = pp.feature_scaling(X_train, transform_arys=(X_train, X_test))