绘图：如何使用热图制作带注释的混淆矩阵？

Question

我喜欢使用 Plotly 来可视化所有内容，我试图通过 Plotly 可视化一个混淆矩阵，这是我的代码：

def plot_confusion_matrix(y_true, y_pred, class_names):

    confusion_matrix = metrics.confusion_matrix(y_true, y_pred)

    confusion_matrix = confusion_matrix.astype(int)

    layout = {

        "title": "Confusion Matrix", 

        "xaxis": {"title": "Predicted value"}, 

        "yaxis": {"title": "Real value"}

    }

    fig = go.Figure(data=go.Heatmap(z=confusion_matrix,

                                    x=class_names,

                                    y=class_names,

                                    hoverongaps=False),

                    layout=layout)

    fig.show()

结果是

如何在相应的单元格中显示数字，而不是像这样悬停

Answer 1

您可以使用带注释的热图来获得以下内容：ff.create_annotated_heatmap()完整代码：import plotly.figure_factory as ffz = [[0.1, 0.3, 0.5, 0.2],&nbsp; &nbsp; &nbsp;[1.0, 0.8, 0.6, 0.1],&nbsp; &nbsp; &nbsp;[0.1, 0.3, 0.6, 0.9],&nbsp; &nbsp; &nbsp;[0.6, 0.4, 0.2, 0.2]]x = ['healthy', 'multiple diseases', 'rust', 'scab']y =&nbsp; ['healthy', 'multiple diseases', 'rust', 'scab']# change each element of z to type string for annotationsz_text = [[str(y) for y in x] for x in z]# set up figure&nbsp;fig = ff.create_annotated_heatmap(z, x=x, y=y, annotation_text=z_text, colorscale='Viridis')# add titlefig.update_layout(title_text='<i><b>Confusion matrix</b></i>',&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; #xaxis = dict(title='x'),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; #yaxis = dict(title='x')&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;)# add custom xaxis titlefig.add_annotation(dict(font=dict(color="black",size=14),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; x=0.5,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; y=-0.15,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; showarrow=False,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; text="Predicted value",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; xref="paper",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; yref="paper"))# add custom yaxis titlefig.add_annotation(dict(font=dict(color="black",size=14),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; x=-0.35,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; y=0.5,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; showarrow=False,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; text="Real value",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; textangle=-90,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; xref="paper",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; yref="paper"))# adjust margins to make room for yaxis titlefig.update_layout(margin=dict(t=50, l=200))# add colorbarfig['data'][0]['showscale'] = Truefig.show()

Answer 2

我发现@vestland的策略是最有用的。但是，与传统的混淆矩阵不同，正确的模型预测是沿着右上角线，而不是左上角。这可以通过反转混淆矩阵的所有索引值来轻松修复，如下所示：import plotly.figure_factory as ffz = [[0.1, 0.3, 0.5, 0.2],&nbsp; &nbsp; &nbsp;[1.0, 0.8, 0.6, 0.1],&nbsp; &nbsp; &nbsp;[0.1, 0.3, 0.6, 0.9],&nbsp; &nbsp; &nbsp;[0.6, 0.4, 0.2, 0.2]]# invert z idx valuesz = z[::-1]x = ['healthy', 'multiple diseases', 'rust', 'scab']y =&nbsp; x[::-1].copy() # invert idx values of x# change each element of z to type string for annotationsz_text = [[str(y) for y in x] for x in z]# set up figure&nbsp;fig = ff.create_annotated_heatmap(z, x=x, y=y, annotation_text=z_text, colorscale='Viridis')# add titlefig.update_layout(title_text='<i><b>Confusion matrix</b></i>',&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; #xaxis = dict(title='x'),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; #yaxis = dict(title='x')&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;)# add custom xaxis titlefig.add_annotation(dict(font=dict(color="black",size=14),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; x=0.5,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; y=-0.15,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; showarrow=False,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; text="Predicted value",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; xref="paper",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; yref="paper"))# add custom yaxis titlefig.add_annotation(dict(font=dict(color="black",size=14),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; x=-0.35,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; y=0.5,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; showarrow=False,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; text="Real value",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; textangle=-90,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; xref="paper",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; yref="paper"))# adjust margins to make room for yaxis titlefig.update_layout(margin=dict(t=50, l=200))# add colorbarfig['data'][0]['showscale'] = Truefig.show()

Answer 3

正如@vestland所说，你可以用情节注释数字。热图可用作任何类型的绘图。这是一个用于从混淆矩阵（基本上只是一个带有数字的2-d向量）绘制热图的代码。def plot_confusion_matrix(cm, labels, title):# cm : confusion matrix list(list)# labels : name of the data list(str)# title : title for the heatmapdata = go.Heatmap(z=cm, y=labels, x=labels)annotations = []for i, row in enumerate(cm):&nbsp; &nbsp; for j, value in enumerate(row):&nbsp; &nbsp; &nbsp; &nbsp; annotations.append(&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; {&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "x": labels[i],&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "y": labels[j],&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "font": {"color": "white"},&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "text": str(value),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "xref": "x1",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "yref": "y1",&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "showarrow": False&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; }&nbsp; &nbsp; &nbsp; &nbsp; )layout = {&nbsp; &nbsp; "title": title,&nbsp; &nbsp; "xaxis": {"title": "Predicted value"},&nbsp; &nbsp; "yaxis": {"title": "Real value"},&nbsp; &nbsp; "annotations": annotations}fig = go.Figure(data=data, layout=layout)return fig