如何在熊猫中的两个变量（列）上写条件

Question

I'm trying to count instances when there was no login, but there was a card-view, and create a new column with the count (or True). I used the conditional statement below and got a key error. Can someone help me figure out what's going on?

import pandas as pd
import numpy as np

sample = pd.DataFrame({ 'Month' : pd.Categorical(["Jan", "Jan", "Feb",  "Feb", "March","Apr", "May"]),
'Name' : pd.Categorical(["Peter", "Meg", "Peter", "Meg", "Meg","Lois", "Lois"]),
'Logins': [1, 1, 1, 1, 1, 1, 0],
'Card': [1, 1, 2, 2, 1, 2, 1]})

sample['LoginNoCard'] = sample['Logins'].where((sample['Logins'] == 0) & (sample['Card'] > 0), sample[1])

The solution I have here is creating a new Data Frame. I'd like to create a new column using a conditional. If Logins == 0 & Card > 0, then 0. If Logins > 0 and Card == 0, then 1. Else NaN.

Answer 1

You can consider using nested np.where() conditions for if Logins == 0 & Card > 0 , then 0 , if Logins > 0 and Card == 0 , then 1 , else NaN .

In [81]: np.where(((sample['Logins']==0) & (sample['Card']>0)), 0,
                    np.where(((sample['Logins']>0) & (sample['Card']==0)), 1,
                    pd.np.nan))
Out[81]: array([ nan,  nan,  nan,  nan,  nan,  nan,   0.])

To assign this to a column, you could

In [82]: sample['LoginNoCard'] = np.where(((sample['Logins']==0) & (sample['Card']>0)), 0,
                                            np.where(((sample['Logins']>0) & (sample['Card']==0)), 1,
                                            pd.np.nan))
In [83]: sample
Out[83]:
   Card  Logins  Month   Name  LoginNoCard
0     1       1    Jan  Peter          NaN
1     1       1    Jan    Meg          NaN
2     2       1    Feb  Peter          NaN
3     2       1    Feb    Meg          NaN
4     1       1  March    Meg          NaN
5     2       1    Apr   Lois          NaN
6     1       0    May   Lois            0