回归问题——Python

1.线性回归

#numpy实现
import numpy as np

#回归系数a，b
a,b=np.dot(np.linalg.inv(np.dot(data_df['feature'].T, data_df['feature'])), np.dot(data_df['feature'].T, data_df['label']))
#预测值
y_predict = part_data_df['feature'] * a + b
#真实值
y_real = part_data_df['label']

#SSE残差平方和
SSE=sum((y_real-y_predict) ** 2)
#SSE2=sum((y_real-y_predict2) ** 2)
#决定系数R2
R2=1 - sum((y_real-y_predict) ** 2)/ sum((part_data_df['label'] - part_data_df['label'].sum()) ** 2)

#sklearn实现
from sklearn.linear_model import LinearRegression
LR = LinearRegression()
LR.fit(part_data_df['feature'].values.reshape(-1,1), part_data_df['label'].values.reshape(-1,1))
y_predict2=LR.predict(part_data_df['feature'].values.reshape(-1,1))

2.逻辑回归

from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix
from sklearn.metrics import precision_score, recall_score, roc_auc_score,f1_score
from sklearn.metrics import roc_curve, roc_auc_score
LR=LogisticRegression()
LR.fit(X=train_x, y=train_y)
LR.score(test_x, test_y)
pred_test_y = LR.predict(test_x)
conf_matrix = confusion_matrix(pred_test_y, test_y)
precision_score(pred_test_y, test_y)
recall_score(pred_test_y, test_y)
f1_score(pred_test_y, test_y)
train_probs = LR.predict_proba(test_x)[:,1]#取第二列的概率，即标签为1的概率
test_probs = LR.predict_proba(test_x)[:,1]
#计算AUC
auc_test = roc_auc_score(test_y, test_probs)
# 计算ROC曲线
fpr, tpr, thresholds = roc_curve(test_y, test_probs, pos_label=1)#自动将probs转换成label
#得到回归系数
coefs = logistic_fit.coef_
#得到截距
intercept = logistic_fit.intercept_