机器学习--数据判断依据 精确度、召回率、调和平均值F1值

精准度(precision)

precision = 正确预测的个数(TP) / 被预测正确的个数(TP+FP)

召回率(recall)

recall = 正确预测的个数(TP)/ 预测个数(FN)

调和平均值 F1-Socre

f1 = 2*精准度 * 召回率 /(精度 * 召回率)

以二分类问题为例

真实\预测 0 1
0 预测negative正确(TN) 预测positive错误(FP)
1 预测positive错误(FN) 预测positive正确(TP)

precision = TP/(TP+FP)
recall = TP/(TP+FN)
求f1_score

from sklearn.metrics import f1_score
f1_score(y_test,y_predict)
# y_test 测试集
# y_predict 预测结果

代码实现上述定义

import numpy as np
from sklearn import datasets

digits = datasets.load_digits()
X = digits['data']
y = digits['target'].copy()

# 手动让digists数据9的数据偏斜
y[digits['target']==9]=1
y[digits['target']!=9]=0

from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split

X_train,X_test,y_train,y_test = train_test_split(X,y,random_state=1)
log_reg = LogisticRegression()
log_reg.fit(X_train,y_train)
log_reg.score(X_test,y_test)
y_log_predict = log_reg.predict(X_test)

# 预测negative正确
def TN(y_true,y_predict):
    return np.sum((y_true==0)&(y_predict==0))
tn = TN(y_test,y_log_predict) # tn = 399

# 预测positive错误
def FP(y_true,y_predict):
    return np.sum((y_true==0)&(y_predict==1))
fp = FP(y_test,y_log_predict) # fp = 3

# 预测negative错误
def FN(y_true,y_predict):
    return np.sum((y_true==1)&(y_predict==0))
fn = FN(y_test,y_log_predict) # fn = 7

# 预测positive正确
def TP(y_true,y_predict):
    return np.sum((y_true==1)&(y_predict==1))
tp = TP(y_test,y_log_predict) # tp = 41

构造混淆矩阵

def confusion_matrix(y_true,y_predict):
    return np.array([
        [TN(y_true,y_predict),FP(y_true,y_predict)],
        [FN(y_true,y_predict),TP(y_true,y_predict)]
    ])
confusion_matrix(y_test,y_log_predict)
"""
output :
array([[399,   3],
       [  7,  41]])
"""

精准率

def precision_score(y_true,y_predict):
    tp = TP(y_true,y_predict)
    fp = FP(y_true,y_predict)
    try:
        return tp/(tp+fp)
    except:
        return 0.0
precision_score(y_test,y_log_predict)
"""
output : 0.9318181818181818
"""

召回率

def recall_score(y_true,y_predict):
    tp = TP(y_true,y_predict)
    fn = FN(y_true,y_predict)
    try:
        return tp/(tp+fn)
    except:
        return 0.0
recall_score(y_test,y_log_predict)
"""
output : 0.8541666666666666
"""

以上为拆分理解 在sklearn中都可以直接求得

# 混淆矩阵
from sklearn.metrics import confusion_matrix
confusion_matrix(y_test,y_log_predict)
"""
output:array([[399,   3],
       [  7,  41]], dtype=int64)
"""
# 精准率
from sklearn.metrics import precision_score
precision_score(y_test,y_log_predict)
"""
output: 0.9318181818181818
"""
# z召回率
from sklearn.metrics import recall_score
recall_score(y_test,y_log_predict)
"""
output: 0.8541666666666666
"""
# classification_report
from sklearn.metrics import classification_report
print(classification_report(y_test,y_log_predict))
"""
output:              precision    recall  f1-score   support

          			0       0.98      0.99      0.99       402
          			1       0.93      0.85      0.89        48

avg / total       			0.98      0.98      0.98       450
"""

你可能感兴趣的:(机器学习)