具有匹配列的多个数据集的相关矩阵热图答案

【问题标题】：Correlation matrix heatmap with multiple datasets that have matching columns具有匹配列的多个数据集的相关矩阵热图
【发布时间】：2017-04-22 12:58:28
【问题描述】：

如果我们有三个数据集：

X = pd.DataFrame({"t":[1,2,3,4,5],"A":[34,12,78,84,26], "B":[54,87,35,25,82], "C":[56,78,0,14,13], "D":[0,23,72,56,14], "E":[78,12,31,0,34]})
Y = pd.DataFrame({"t":[1,2,3,4,5],"A":[45,24,65,65,65], "B":[45,87,65,52,12], "C":[98,52,32,32,12], "D":[0,23,1,365,53], "E":[24,12,65,3,65]})
Z = pd.DataFrame({"t":[1,2,3,4,5],"A":[14,96,25,2,25], "B":[47,7,5,58,34], "C":[85,45,65,53,53], "D":[3,35,12,56,236], "E":[68,10,45,46,85]})

其中“t”是一个索引。

如何输出类似于 seaborn 示例的相关矩阵热图：

只是轴看起来像这样：

【问题讨论】：

标签： python pandas numpy correlation seaborn

【解决方案1】：

X = pd.DataFrame({"t":[1,2,3,4,5],"A":[34,12,78,84,26], "B":[54,87,35,25,82], "C":[56,78,0,14,13], "D":[0,23,72,56,14], "E":[78,12,31,0,34]})
Y = pd.DataFrame({"t":[1,2,3,4,5],"A":[45,24,65,65,65], "B":[45,87,65,52,12], "C":[98,52,32,32,12], "D":[0,23,1,365,53], "E":[24,12,65,3,65]})
Z = pd.DataFrame({"t":[1,2,3,4,5],"A":[14,96,25,2,25], "B":[47,7,5,58,34], "C":[85,45,65,53,53], "D":[3,35,12,56,236], "E":[68,10,45,46,85]})


catted = pd.concat([d.set_index('t') for d in [X, Y, Z]], axis=1, keys=['X', 'Y', 'Z'])
catted = catted.rename_axis(['Source', 'Column'], axis=1)

corrmat = catted.corr()

f, ax = plt.subplots()

sns.heatmap(corrmat, vmax=.8, square=True)

sources = corrmat.columns.get_level_values(0)
for i, source in enumerate(sources):
    if i and source != sources[i - 1]:
        ax.axhline(len(sources) - i, c="w")
        ax.axvline(i, c="w")
f.tight_layout()

对评论的回应：
我更改了X、Y 和Z 中的t 列

X = pd.DataFrame({"t":[1,2,3,4,5],"A":[34,12,78,84,26], "B":[54,87,35,25,82], "C":[56,78,0,14,13], "D":[0,23,72,56,14], "E":[78,12,31,0,34]})
Y = pd.DataFrame({"t":[6,7,8,9,10],"A":[45,24,65,65,65], "B":[45,87,65,52,12], "C":[98,52,32,32,12], "D":[0,23,1,365,53], "E":[24,12,65,3,65]})
Z = pd.DataFrame({"t":[11,12,13,14,15],"A":[14,96,25,2,25], "B":[47,7,5,58,34], "C":[85,45,65,53,53], "D":[3,35,12,56,236], "E":[68,10,45,46,85]})


catted = pd.concat([d.set_index('t') for d in [X, Y, Z]], axis=1, keys=['X', 'Y', 'Z'])
catted = catted.rename_axis(['Source', 'Column'], axis=1)

corrmat = catted.corr()

f, ax = plt.subplots()

sns.heatmap(corrmat, vmax=.8, square=True)

sources = corrmat.columns.get_level_values(0)
for i, source in enumerate(sources):
    if i and source != sources[i - 1]:
        ax.axhline(len(sources) - i, c="w")
        ax.axvline(i, c="w")
f.tight_layout()

现在又来了，不过我reset_index

X = pd.DataFrame({"t":[1,2,3,4,5],"A":[34,12,78,84,26], "B":[54,87,35,25,82], "C":[56,78,0,14,13], "D":[0,23,72,56,14], "E":[78,12,31,0,34]})
Y = pd.DataFrame({"t":[6,7,8,9,10],"A":[45,24,65,65,65], "B":[45,87,65,52,12], "C":[98,52,32,32,12], "D":[0,23,1,365,53], "E":[24,12,65,3,65]})
Z = pd.DataFrame({"t":[11,12,13,14,15],"A":[14,96,25,2,25], "B":[47,7,5,58,34], "C":[85,45,65,53,53], "D":[3,35,12,56,236], "E":[68,10,45,46,85]})


catted = pd.concat([d.reset_index(drop=True) for d in [X, Y, Z]], axis=1, keys=['X', 'Y', 'Z'])
catted = catted.rename_axis(['Source', 'Column'], axis=1)

corrmat = catted.corr()

f, ax = plt.subplots()

sns.heatmap(corrmat, vmax=.8, square=True)

sources = corrmat.columns.get_level_values(0)
for i, source in enumerate(sources):
    if i and source != sources[i - 1]:
        ax.axhline(len(sources) - i, c="w")
        ax.axvline(i, c="w")
f.tight_layout()

【讨论】：

你知道为什么当我将相关矩阵应用于更大的数据时，它只显示对角线正方形吗？见图片：i.imgur.com/hLorwN2.png
我怀疑t 列未与X、Y 和Z 对齐
在玩了之后，我发现我在这些块中的相关性非常小，以至于看起来这些方块中什么都没有。我改变了我的规模，现在它是完美的。感谢@piRSquared 的帮助。