机器学习：分类模型的评估精确率Presicion和召回率Recall

estimator.score()
准确率：预测结果正确的百分比

混淆矩阵

预测结果Predicted Condition
正确标记 True Condition

预测结果	正例	假例
真实 正例	真正例TP	伪反例FN
结果 假例	伪正例FP	真反例TN

T True
F False
P Positive
N Negative

精确率 Presicion
预测结果为正中真实为正的比例（查的准）

召回率 Recall
真实为正中预测结果为正的比例（查的全，对正样本的区分能力）

F1-score 模型的稳健性
F1=(2TP)/(2TP + FN + FP)
= (2 x Precision x Recall)/(Precision + Recall)

代码示例

from sklearn.datasets import fetch_20newsgroups
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.model_selection import train_test_split
from sklearn.naive_bayes import MultinomialNB
from sklearn.metrics import classification_report
import ssl

ssl._create_default_https_context = ssl._create_unverified_context

# 如果获取不到就下载
data = fetch_20newsgroups(subset="all")

# 数据分割
X_train, X_test, y_train, y_test = train_test_split(
    data.data, data.target, test_size=0.33, random_state=42
)

# 特征抽取
tfidf = TfidfVectorizer()

# 以训练集中的词列表对每篇文章做重要性统计
X_train = tfidf.fit_transform(X_train)
print(tfidf.get_feature_names())

X_test = tfidf.transform(X_test)

# 朴素贝叶斯算法预测，alpha是拉普拉斯平滑系数
mlt = MultinomialNB(alpha=1.0)
mlt.fit(X_train, y_train)
y_predict = mlt.predict(X_test)
score = mlt.score(X_test, y_test)
print("socre: {}".format(score))
# socre: 0.83

# 分类报告
print(classification_report(y_test, y_predict, target_names=data.target_names))
"""
                          precision    recall  f1-score   support

             alt.atheism       0.86      0.71      0.78       260
           comp.graphics       0.86      0.77      0.81       321
 comp.os.ms-windows.misc       0.82      0.83      0.82       314
 ...
             avg / total       0.87      0.83      0.83      6220
"""