In [ ]:

    

# Sample Generation

%matplotlib inline

import matplotlib.pyplot as plt
import numpy as np

from sklearn.datasets.samples_generator import make_blobs


# X为样本特征，y为样本簇类别，共1000个样本，每个样本2个特征，共2个簇
x_all, y_all = make_blobs(n_samples=1000, n_features=2, 
                          centers=[[5,5], [6,6]],
                          cluster_std=[0.3, 0.4],
                          random_state=9)

plt.figure()
colors = ['r', 'b']
for y, c in zip(np.unique(y_all), colors):
    plt.scatter(X_all[y_all==y, 0], x_all[y_all==y, 1], c=c, label=y, marker='o')


    

In [ ]:

    

# Explain decision_function, predict_proba and predict

from sklearn.datasets.samples_generator import make_blobs
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
import pandas as pd
import numpy as np


def sigmoid_array(x):                                        
    return 1 / (1 + np.exp(-x))

    
x_all, y_all = make_blobs(n_samples=1000, n_features=2, 
                          centers=[[5,5], [6,6]],
                          cluster_std=[1, 1],
                          random_state=9)

x_train, x_test, y_train, y_test = train_test_split(x_all, y_all, random_state=1)

clf = LogisticRegression(C=1, intercept_scaling=1, random_state=1)
clf.fit(x_train, y_train)

print(clf.classes_)
print(clf.coef_)
print(clf.intercept_)
print(clf.n_iter_)
print(clf.densify())
print(clf.get_params(deep=True))

y_pred = clf.predict(x_test)
y_pred_proba = clf.predict_proba(x_test)
y_confidence = clf.decision_function(x_test)
y_sigmoid = sigmoid_array(y_confidence)

result = pd.DataFrame({'y': y_test,
                       'y_pred': y_pred,
                       'y_pred_proba': y_pred_proba[:, 1],
                       'y_confidence': y_confidence,
                       'y_sigmoid': y_sigmoid})
print(result)


    

In [ ]:

    

# Explain how to calculate decision_function

from sklearn.datasets.samples_generator import make_blobs
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
import pandas as pd
import numpy as np


def sigmoid_array(x):                                        
    return 1 / (1 + np.exp(-x))

    
x_all, y_all = make_blobs(n_samples=1000, n_features=2, 
                          centers=[[5,5], [6,6]],
                          cluster_std=[1, 1],
                          random_state=9)

x_train, x_test, y_train, y_test = train_test_split(x_all, y_all, random_state=1)

clf = LogisticRegression(C=1, intercept_scaling=1, random_state=1)
clf.fit(x_train, y_train)

print(clf.classes_)
print(clf.coef_)
print(clf.intercept_)

def func(x):
    return x[0] * clf.coef_[0][0] + x[1] * clf.coef_[0][1] + clf.intercept_

y_calc = np.apply_along_axis(func, axis=1, arr=x_test)
y_confidence = clf.decision_function(x_test)

result = pd.DataFrame({'y_calc': y_calc[:, 0], 'y_confidence': y_confidence})
print(result)

print('Finished')


    

In [ ]:

    

# Example for intercept_scaling in LogisticRegression

from sklearn.datasets.samples_generator import make_blobs
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
import pandas as pd
import numpy as np


x_all, y_all = make_blobs(n_samples=1000, n_features=2, 
                          centers=[[5,5], [6,6]],
                          cluster_std=[2, 2],
                          random_state=9)

x_train, x_test, y_train, y_test = train_test_split(x_all, y_all, random_state=1)


sc_list = np.linspace(0.1, 1, 10)
for sc in sc_list:
    print('intercept_scaling = {}'.format(sc))
    clf = LogisticRegression(C=1, intercept_scaling=sc, random_state=1)
    clf.fit(x_train, y_train)
    # print(classification_report(y_train, clf.predict(x_train)))
    print(classification_report(y_test, clf.predict(x_test)))


    

In [ ]:

    

# Example for Prediction Threshold

from sklearn.datasets.samples_generator import make_blobs
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
import pandas as pd
import numpy as np


x_all, y_all = make_blobs(n_samples=1000, n_features=2, 
                          centers=[[5,5], [6,6]],
                          cluster_std=[2, 2],
                          random_state=9)

x_train, x_test, y_train, y_test = train_test_split(x_all, y_all, random_state=1)

clf = LogisticRegression(C=1, intercept_scaling=1, random_state=1)
clf.fit(x_train, y_train)

print(classification_report(y_train, clf.predict(x_train)))
print(classification_report(y_test, clf.predict(x_test)))

pred_proba_train = clf.predict_proba(x_train)[:, 1]
pred_proba_test = clf.predict_proba(x_test)[:, 1]

print(classification_report(y_train, (pred_proba_train > 0.6)))
print(classification_report(y_test, (pred_proba_test > 0.6)))


    

In [ ]:

    

# Grid Search

from sklearn.datasets.samples_generator import make_blobs
from sklearn.model_selection import train_test_split
from sklearn.model_selection import GridSearchCV
from sklearn.metrics import classification_report
from sklearn import svm


x_all, y_all = make_blobs(n_samples=1000, n_features=2, 
                          centers=[[5,5], [6,6]],
                          cluster_std=[2, 2],
                          random_state=9)

x_train, x_test, y_train, y_test = train_test_split(x_all, y_all, test_size=0.2, random_state=0)

tuned_parameters = [{'C': [0.001, 0.01, 0.1, 1, 10, 100, 1000],
                     'intercept_scaling': [0.1, 0.5, 1, 5, 10]}]

# Tuning hyper-parameters
clf = GridSearchCV(LogisticRegression(),tuned_parameters, cv=5)
clf.fit(x_train, y_train)

print("Best score found on test set:")
print(clf.best_score_)
print()

print("Best score found on test set:")
print(clf.best_estimator_)
print()

print("Best parameters set found on test set:")
print(clf.best_params_)
print(clf.best_estimator_.get_params())
print()

print("Grid scores on test set:")
means = clf.cv_results_['mean_test_score']
stds = clf.cv_results_['std_test_score']
params = clf.cv_results_['params']
for mean, std, params in zip(means, stds, params):
    print("%0.3f (+/-%0.03f) for %r" % (mean, std * 2, params))
print()

print("Detailed classification report:")
y_true, y_pred = y_test, clf.predict(x_test)
print('Score of best_estimator_: {}'.format(clf.score(x_test, y_test)))
print(classification_report(y_true, y_pred))