使用pandas.DataFrame.pivot（不聚合values=）或pandas.DataFrame.pivot_table（聚合values=）将 Dataframe 从长格式重新调整为宽格式。索引将位于y轴上，列将位于x轴上。有关概述，请参见Reshaping and pivot tables。

In [96]: result
Out[96]: 
MutProb               0.001      0.005      0.010     0.050     0.100
SymmetricDivision                                                    
0.2               -6.146121  -8.571063  -9.784686 -6.051482 -0.964818
0.4               -6.473629  -8.936463  -9.455776 -6.885229 -0.652147
0.6               -6.760559  -9.292469  -9.551801 -6.621639 -0.392256
0.8               -7.196407  -9.544065 -10.536340 -6.996394 -0.722602
1.0               -8.027475 -10.502450 -11.408114 -9.175349 -4.180864

然后，您可以将二维数组（或DataFrame）传递给seaborn.heatmap或plt.pcolor：

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# load the sample data
df = pd.DataFrame({'MutProb': [0.1,
  0.05, 0.01, 0.005, 0.001, 0.1, 0.05, 0.01, 0.005, 0.001, 0.1, 0.05, 0.01, 0.005, 0.001, 0.1, 0.05, 0.01, 0.005, 0.001, 0.1, 0.05, 0.01, 0.005, 0.001], 'SymmetricDivision': [1.0, 1.0, 1.0, 1.0, 1.0, 0.8, 0.8, 0.8, 0.8, 0.8, 0.6, 0.6, 0.6, 0.6, 0.6, 0.4, 0.4, 0.4, 0.4, 0.4, 0.2, 0.2, 0.2, 0.2, 0.2], 'test': ['sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule'], 'value': [-4.1808639999999997, -9.1753490000000006, -11.408113999999999, -10.50245, -8.0274750000000008, -0.72260200000000008, -6.9963940000000004, -10.536339999999999, -9.5440649999999998, -7.1964070000000007, -0.39225599999999999, -6.6216390000000001, -9.5518009999999993, -9.2924690000000005, -6.7605589999999998, -0.65214700000000003, -6.8852289999999989, -9.4557760000000002, -8.9364629999999998, -6.4736289999999999, -0.96481800000000006, -6.051482, -9.7846860000000007, -8.5710630000000005, -6.1461209999999999]})

# pivot the dataframe from long to wide form
result = df.pivot(index='SymmetricDivision', columns='MutProb', values='value')

sns.heatmap(result, annot=True, fmt="g", cmap='viridis')
plt.show()

收益率

使用.grouby()和.unstack()的另一个选项

df_m = df.groupby(["SymmetricDivision", "MutProb"])
          .size()
          .unstack(level=0)

sns.heatmap(df_m)

我必须创建一个热图来表示DecisionTreeClassifier模型的超参数调优结果，超参数为['max_depth'，'min_samples_split']。表示模型精度的指标为auc_roc_score。
我们将使用seaborn heatmap为sns.heatmap（）创建一个数据集。
我们使用pandas.pivot_table（）将DataFrame转换为pandas。在制作热图之前要做的操作之一是使用Pandas的转换功能来重塑数据以制作热图。

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt[![enter image description here][2]][2]

max_depth = [1, 5, 10, 50]
min_samples_split = [5, 10, 100, 500]

train_set_auc = [0.5478688747761202, 0.5478688747761202, 0.5478688747761202, 
                 0.5478688747761202, 0.6384311073066989, 0.6383831998001249, 
                 0.6382879025310534, 0.6381476488966438, 0.7099806625823346, 
                 0.7090830674744466, 0.7035862433327912, 0.6987068807767094, 
                 0.9203751110844185, 0.9140516357169505, 0.8909876822536001, 
                 0.845991656101182]

# Now, let me explain the approach
# 1. we will use pandas.pivot_table()[seaborn heatmap tutorial][1] [pandas pivot_table()][1] to create a 
# dataset for sns.heatmap()

# convert train_set_auc into 4X4 np array

def return_dataframe(lst, max_depth, min_samples):
    """ this function will take a list and convert it into a 
        4X4 numpy array. Further, it will create an empty dataframe 
        with columns= min_samples_split and indices as max_depth. Then 
        it will fill the entries with values from the numpy array
   """
    arr = []
    idx = 0

    for idx in range(0, len(lst), 4):
        arr.append(np.array(lst[idx:idx+4]))

    arr = np.array(arr)

    df = pd.DataFrame(data=arr, index=min_samples, columns=max_depth)

   return df

train_set1 = return_dataframe(lst=train_set1_auc, max_depth=max_depth, min_samples=min_samples_split)
fig, ax = plt.subplots()

ax = sns.heatmap(train_set1, annot=True, fmt='g', cmap='viridis')
plt.xlabel("Your x-label")
plt.ylabel("Your y-label")
plt.show()

我在编辑中添加了plt.xlabel（）& plt.ylabel（）。很抱歉，现在显示在下面的图像中。顺便说一句，这个可以工作。如果你还没有尝试一次

使用pd.DataFrame.pivot()透视数据后，还可以使用pd.DataFrame.style创建热图。

加载数据

import pandas as pd

# load the sample data
df = pd.DataFrame({'MutProb': [0.1,
  0.05, 0.01, 0.005, 0.001, 0.1, 0.05, 0.01, 0.005, 0.001, 0.1, 0.05, 0.01, 0.005, 0.001, 0.1, 0.05, 0.01, 0.005, 0.001, 0.1, 0.05, 0.01, 0.005, 0.001], 'SymmetricDivision': [1.0, 1.0, 1.0, 1.0, 1.0, 0.8, 0.8, 0.8, 0.8, 0.8, 0.6, 0.6, 0.6, 0.6, 0.6, 0.4, 0.4, 0.4, 0.4, 0.4, 0.2, 0.2, 0.2, 0.2, 0.2], 'test': ['sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule', 'sackin_yule'], 'value': [-4.1808639999999997, -9.1753490000000006, -11.408113999999999, -10.50245, -8.0274750000000008, -0.72260200000000008, -6.9963940000000004, -10.536339999999999, -9.5440649999999998, -7.1964070000000007, -0.39225599999999999, -6.6216390000000001, -9.5518009999999993, -9.2924690000000005, -6.7605589999999998, -0.65214700000000003, -6.8852289999999989, -9.4557760000000002, -8.9364629999999998, -6.4736289999999999, -0.96481800000000006, -6.051482, -9.7846860000000007, -8.5710630000000005, -6.1461209999999999]})

透视数据

pivot_table_df = df.pivot(index='SymmetricDivision', columns='MutProb', values='value')

创建热图

axis=None表示使用表中的所有值

pivot_table_df.style.background_gradient(axis=None)

创建内嵌条形图

pivot_table_df.style.bar(axis=None)