将SciPy stats.contingency应用于DataFrame以求出数据的临界值和p值时出错

Question

我试图检查DF下面的性别与目的地国家列之间的关系。

              id    gender  country_destination
   2    4ft3gnwmtx  FEMALE  US
   6    lsw9q7uk0j  FEMALE  US
   7    0d01nltbrs  FEMALE  US
   8    a1vcnhxeij  FEMALE  US
  10    yuuqmid2rp  FEMALE  US

为了应用stat.chi2_contingency和获取p值，我将这个表转换为pivot_table，如下所示：

       AU   CA  DE  ES  FR  GB  IT  NL  PT  US
gender                                      
FEMALE 207  455 358 853 1962 88 1091 254 78 22694
MALE   188  477 416 677 1335 682 699 278 69 19457

但是，当我执行stats.chi2_contingency(df_contingency)时，我会出现以下错误

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-161-fee39cdf166f> in <module>()
----> 1 stats.chi2_contingency(df_contingency)

/usr/local/lib/python3.6/dist-packages/scipy/stats/contingency.py in 
chi2_contingency(observed, correction, lambda_)
    243     observed = np.asarray(observed)
--> 244     if np.any(observed < 0):
    245         raise ValueError("All values in `observed` must be nonnegative.")
    246     if observed.size == 0:

TypeError: '<' not supported between instances of 'str' and 'int'

请帮我弄清楚我哪里出了问题？

Answer 1

我猜你做了这样的事：

import pandas as pd
import numpy as np
from scipy.stats import chi2_contingency

np.random.seed(111)

dat = pd.DataFrame({'gender':np.random.choice(['MALE','FEMALE'],100),
                  'country_destination':np.random.choice(['AU','CA','DE','ES','FR'],100)})

pt = dat.pivot_table(columns='country_destination',index='gender',
values='country_destination',aggfunc=len)

country_destination AU  CA  DE  ES  FR
gender                  
FEMALE  6   12  12  9   15
MALE    8   9   12  4   13

这项工程的意外事故：

chi2_contingency(pt)

如果你做交叉表而不是枢轴，你也会：

chi2_contingency(pd.crosstab(dat['gender'],dat['country_destination']))

最有可能的情况是，您需要提供指向应急表的计数。