Python 多个分类变量之间的相关性（Pandas）
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/48035381/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Correlation among multiple categorical variables (Pandas)
提问by zar3bski
I have a data set made of 22 categorical variables(non-ordered). I would like to visualize their correlation in a nice heatmap. Since the Pandas built-in function
我有一个由 22 个分类变量（无序）组成的数据集。我想在一个不错的热图中可视化它们的相关性。由于 Pandas 内置函数
DataFrame.corr(method='pearson', min_periods=1)
only implement correlation coefficients for numerical variables (Pearson, Kendall, Spearman), I have to aggregate it myself to perform a chi-square or something like it and I am not quite sure which function use to do it in one elegant step(rather than iterating through all the cat1*cat2 pairs). To be clear, this is what I would like to end up with (a dataframe):
只实现数值变量（Pearson、Kendall、Spearman）的相关系数，我必须自己聚合它来执行卡方或类似的东西，我不太确定使用哪个函数在一个优雅的步骤中完成它（而不是遍历所有 cat1*cat2 对）。需要明确的是，这就是我想要的结果（一个数据框）：
cat1 cat2 cat3
cat1| coef coef coef
cat2| coef coef coef
cat3| coef coef coef
Any ideas with pd.pivot_tableor something in the same vein?
任何关于pd.pivot_table或类似内容的想法？
thanks in advance D.
提前致谢。
采纳答案by YOBEN_S
You can using pd.factorize
您可以使用 pd.factorize
df.apply(lambda x : pd.factorize(x)[0]).corr(method='pearson', min_periods=1)
Out[32]:
a c d
a 1.0 1.0 1.0
c 1.0 1.0 1.0
d 1.0 1.0 1.0
Data input
数据输入
df=pd.DataFrame({'a':['a','b','c'],'c':['a','b','c'],'d':['a','b','c']})
Update
更新
from scipy.stats import chisquare
df=df.apply(lambda x : pd.factorize(x)[0])+1
pd.DataFrame([chisquare(df[x].values,f_exp=df.values.T,axis=1)[0] for x in df])
Out[123]:
0 1 2 3
0 0.0 0.0 0.0 0.0
1 0.0 0.0 0.0 0.0
2 0.0 0.0 0.0 0.0
3 0.0 0.0 0.0 0.0
df=pd.DataFrame({'a':['a','d','c'],'c':['a','b','c'],'d':['a','b','c'],'e':['a','b','c']})
回答by Shashwat Tiwary
Found a nice and clean solution in this post. It's not one steps but provides what is required.Post on correlation for categorical variables
在这篇文章中找到了一个漂亮而干净的解决方案。这不是一个步骤，而是提供了所需的内容。发布分类变量的相关性
回答by zar3bski
Turns out, the only solution I found is to iterate trough all the factor*factor pairs.
事实证明，我找到的唯一解决方案是遍历所有因子 * 因子对。
factors_paired = [(i,j) for i in df.columns.values for j in df.columns.values]
chi2, p_values =[], []
for f in factors_paired:
if f[0] != f[1]:
chitest = chi2_contingency(pd.crosstab(df[f[0]], df[f[1]]))
chi2.append(chitest[0])
p_values.append(chitest[1])
else: # for same factor pair
chi2.append(0)
p_values.append(0)
chi2 = np.array(chi2).reshape((23,23)) # shape it as a matrix
chi2 = pd.DataFrame(chi2, index=df.columns.values, columns=df.columns.values) # then a df for convenience
