机器学习模型的准确度测试的最佳计算方法是什么

Question

我想知道计算精度测试的正确方法

This is a code for training model then calculate acuracy for testing. I used three method for calculate accuracy:

• 1-准确度评分（）
• 2-model.score

• 3-u矩阵

  SVMclassifier = SVC()
  SVMclassifier.fit(X_train, Y_train)
  SVMpred=SVMclassifier.predict(X_test)
  print('accuracy score for test:',accuracy_score(Y_test, SVMpred)*100)
  SVMres = model.score(X_test, Y_test)
  print("Accuracy res: %.3f%%" % (SVMres*100.0))
  SVMconf = metrics.confusion_matrix(Y_test,SVMpred)
  TP = conf[1, 1]
  TN = conf[0, 0]
  FP = conf[0, 1]
  FN = conf[1, 0]
  SVMacc = (TP+TN)/(TP+TN+FP+FN)
  SVMspec= TN / (TN + FP)
  SVMsens = TP / float(FN + TP)
  print('confusion_matrix',SVMconf)
  print('accuracy ', SVMacc*100.0)
  

The output accuracy:

1. 准确度评分（）：98.456

2. 模型分数：98.45

3. 混淆矩阵：97.2