Why can not pass the validation of type of the series-CodePudding

I do not know why it can not pass the validation of each variable.

marvel_df = rate_df.loc[rate_df['Company']== "Marvel"]
mean_marvel =marvel_df[['Rate']].mean()
std_marvel =marvel_df[['Rate']].std()
n_marvel = marvel_df[['Rate']].count()

dc_df = rate_df.loc[rate_df['Company']== "DC"]
mean_dc =dc_df[['Rate']].mean()
std_dc =dc_df[['Rate']].std()
n_dc =dc_df[['Rate']].count()

# Validation
assert n_marvel == 23
assert n_dc == 16
assert np.ndim(mean_marvel) == np.ndim(mean_dc) == np.ndim(std_marvel) == np.ndim(std_dc) == np.ndim(n_marvel) == np.ndim(n_dc)

CodePudding user response：

Try it with this cleaned up code:

marvel_df = rate_df[rate_df['Company'] == "Marvel"]
mean_marvel = marvel_df['Rate'].mean()
std_marvel = marvel_df['Rate'].std()
n_marvel = marvel_df['Rate'].count()

dc_df = rate_df[rate_df['Company'] == "DC"]
mean_dc = dc_df['Rate'].mean()
std_dc = dc_df['Rate'].std()
n_dc = dc_df['Rate'].count()

CodePudding user response：

marvel_df[['Rate']].count() gives you a pd.Series with one element, namely the count of the non-NaN values in the Rate column of marvel_df. n_marvel == 23 returns a pd.Series of boolean values, again with only one entry - but the truth value of a Series is ambiguous.