マシューズ相関係数(MCC）のモデル評価

import numpy as np

from sklearn.datasets import make_classification

from sklearn.model_selection import train_test_split

from sklearn.linear_model import LogisticRegression

from sklearn.metrics import roc_auc_score

from sklearn.metrics import matthews_corrcoef

from sklearn.metrics import accuracy_score

from sklearn.metrics import f1_score

from sklearn.metrics import confusion_matrix

# サンプルデータの生成

# 1000 samples、5(infomative) + 2(redundant) + 13(independent) =  20 feature のデータを生成

dat = make_classification(n_samples=10000, n_features=20, n_informative=5, n_redundant=2, n_classes=2, n_clusters_per_class=10,weights=[0.9, 0.1])

X = dat[0]

y = dat[1]

# 学習用とテスト用データの分割

# 80%を学習、20%をテストに利用する

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=123)

# 学習モデルの構築とパフォーマンス評価

# ロジスティック回帰

clf = LogisticRegression()

clf.fit(X_train, y_train)

print (roc_auc_score(y_test, clf.predict_proba(X_test)[:,1]))