笔试-混淆矩阵-召回率

已知一个分类器在测试集上的混淆矩阵如下表所示，该分类器对各个类别的召回率为：

M = [[69, 3, 13, 11, 4],
     [14, 70, 3, 7,  6],
     [3,  3,  80, 9, 5],
     [3,  6,  9,  76,4],
     [8, 11, 4, 23, 54]]

n = len(M)
for i in range(len(M[0])):
    rowsum, colsum = sum(M[i]), sum(M[r][i] for r in range(n))
    try:
        print('precision: %s' % (M[i][i]/float(colsum)), 'recall: %s' % (M[i][i]/float(rowsum)))
    except ZeroDivisionError:
        print('precision: %s' % 0, 'recall: %s' %0)

结果为：

precision: 0.711340206185567 recall: 0.69
precision: 0.7526881720430108 recall: 0.7
precision: 0.7339449541284404 recall: 0.8
precision: 0.6031746031746031 recall: 0.7755102040816326
precision: 0.7397260273972602 recall: 0.54