confusion matrix 混淆矩阵绘制

淡淡的烟草味﹌ 2022-05-12 02:50 297阅读 0赞

### 原理 ###

横轴一般是predict label，纵轴是ground truth label，对角线是预测正确的概率或个数

### 绘制 ###

sklearn 中confusion\_matrix函数的使用：  
`sklearn.metrics.confusion_matrix(y_true, y_pred, labels=None, sample_weight=None)`  
输入就是类别序号，不能是onehotlabel，如果是onehot label 用`y_pred = np.argmax(y_pred,axis= -1)+1` 转换  
y\_true : array, shape = \[n\_samples\] Ground truth (correct) target values.  
y\_pred : array, shape = \[n\_samples\] Estimated targets as returned by a classifier.

from sklearn.metrics import confusion_matrix
    y_true=[2,1,0,1,2,0]
    y_pred=[2,0,0,1,2,1]
    C=confusion_matrix(y_true, y_pred)

![这里写图片描述][70]

纵轴是y\_true, 横轴是y\_pred

##### 一　、带有acc的confusion matrix #####

#label参数`
    labels = ['drink water','pour water','call','play phone','backpack ','tidy bkpk','sitting','move chair','take out wallet','take money','mop','sweep']
    def plot_confusion_matrix(cm, title='Confusion Matrix', cmap = plt.cm.binary):
        plt.imshow(cm, interpolation='nearest', cmap=cmap)
        plt.title(title)
        plt.colorbar()
        xlocations = np.array(range(len(labels))) # （1，c）类别个数作为位置，
        plt.xticks(xlocations, labels, rotation=60) # 将x坐标的label旋转60度 ，每个位置上放好对应的label
        plt.yticks(xlocations, labels)
        plt.ylabel('True label')
        plt.xlabel('Predicted label')
        plt.savefig(model_name + '_'+setting+'.png', format='png') # 存储图片
    
    def draw_matrix_acc(y_pred,y_true):
        tick_marks = np.array(range(len(labels))) + 0.5   
        ind_array = np.arange(len(labels))
        cm = confusion_matrix(y_true, y_pred) # 求解confusion matrix 
        print cm
        np.set_printoptions(precision=2) 
        cm_normalized = cm.astype('float')/cm.sum(axis=1)[:, np.newaxis]
        print cm_normalized
        plt.figure(figsize=(12,8), dpi=120)
    
        x, y = np.meshgrid(ind_array, ind_array)
        for x_val, y_val in zip(x.flatten(), y.flatten()):
            c = cm_normalized[y_val][x_val]
            if (c > 0.01):
                plt.text(x_val, y_val, "%0.2f" %(c,), color='red', fontsize=7, va='center', ha='center')
        #offset the tick
        plt.gca().set_xticks(tick_marks, minor=True)
        plt.gca().set_yticks(tick_marks, minor=True)
        plt.gca().xaxis.set_ticks_position('none')
        plt.gca().yaxis.set_ticks_position('none')
        plt.grid(True, which='minor', linestyle='-')
        plt.gcf().subplots_adjust(bottom=0.15)
    
        plot_confusion_matrix(cm_normalized, title=model_name + '_'+setting)
        #show confusion matrix
        plt.show()
    # 调用方法，
        draw_matrix_acc(y_pred_list,y_true_list)

### 参考 ###

[如何用python画好confusion matrix][python_confusion matrix]  
[这个超强][Link 1]  
[有数值的版本][python_confusion matrix]

[70]: /images/20220512/1c355a229f62442489982ed16d3c22c3.png
[python_confusion matrix]: https://blog.csdn.net/guoyilin/article/details/42047615
[Link 1]: https://blog.csdn.net/epsil/article/details/9171527