Machine Learning Confusion Matrix

The role of the confusion matrix : an indicator for evaluating the learning effect of machine learning. confusion matrix is ​​to evaluate the accuracy of a classification.

Definition of confusion matrix: For confusion matrix C , C (i,j)  indicates that the true classification is theiclass, but the predicted value is thetotal number of observationsj

For a binary classification problem (assuming 1 is the positive class):

C (0,0) represents the true negative class TN (true negative)	C (0,1) represents the false positive class FP (false negative)
C (1,0) represents the false negative class FP (false negative)	C (1,1) represents the true class TP (true positive)

Implementation of confusion matrix:

    By computing the confusion matrix using the scikit-learn module,

  sklearn.metrics.confusion_matrix(y_true, y_pred, labels=None, sample_weight=None).

    The parameters are explained below: (Note: the array type in the table corresponds to the list list() type in python)  

Parameters: parameter:

y_true : array type of  length n_samples               represents the true classification. Ground truth (correct) target values. y_pred  : array type of length n_samples                Represents the predicted classification value for n examples by the classifier. Estimated targets as returned by a classifier. labels  : [optional parameter], an array type of length n_classes ( n_classes represents the number of categories/labels )                This parameter is in the form of a list of labels, which affects the meaning of the row and column representation of the output confusion matrix C, and which type of value is output first .                  This matrix may be used to rearrange or select a subset to compute the confusion matrix.                If this parameter is not given (i.e. None), then  the value that appears at least once in y_true or  y_predwill be used for the sort order                  Notes: 1. The columns of the confusion matrix represent the order of the categories 2. The values ​​appearing in  or  represent the class labels . y_truey_pred                List of labels to index the matrix. This may be used to reorder or select a subset of labels.                 If none is given, those that appear at least once in y_true or y_pred are used in sorted order. sample_weight : 类似长度为n_samples的数组形式，可选参数。                              样例的权重向量。Sample weights.

Returns: 返回值：

C : array, shape = [n_classes, n_classes] 返回混淆矩阵(Confusion matrix)C，shape=(n_classes, n_classes)。

混淆矩阵的性质：

        矩阵的对角线表示正确分类的个数，非对角线的点是分类错误的情况。