Pandas ".loc" method not working as expected-CodePudding

Why is this working?

import pandas as pd

numbers = {'mynumbers': [51, 52, 53, 54, 55]}
df = pd.DataFrame(numbers, columns =['mynumbers'])
df.loc[df['mynumbers'] <= 53, 'mynumbers'] = 'True'
print (df)

Output:

  mynumbers
0      True
1      True
2      True
3     False
4     False

But this returns an error:

import pandas as pd

numbers = {'mynumbers': [51, 52, 53, 54, 55]}
df = pd.DataFrame(numbers, columns =['mynumbers'])
print(df.loc[df['mynumbers']])

If in the first case I can use the "df.loc[df['mynumbers']]" statement as a conditional to compare values, why do I get an error when I simply try to print out the statement alone?

I understand that the index values that I pass into the .loc method yield a key error because there is no such key exist, but I do not understand that why does it works in the first instance?

CodePudding user response：

When you do df['mynumbers'] <= 53 you use a boolean indexer, that is a series that has the same index as df and either True or False as values:

>>> df['mynumbers'] <= 53
0     True
1     True
2     True
3    False
4    False
Name: mynumbers, dtype: bool

This can be passed to df.loc[] or df[]:

>>> df[df['mynumbers'] <= 53]
   mynumbers
0         51
1         52
2         53
>>> df.loc[df['mynumbers'] <= 53, :]
   mynumbers
0         51
1         52
2         53

The other way to use df.loc[] is to pass in index values:

>>> df.loc[df.index]
   mynumbers
0         51
1         52
2         53
3         54
4         55
>>> df.loc[df.index[3:]]
   mynumbers
3         54
4         55
>>> df.loc[[1, 2]]
   mynumbers
1         52
2         53

However when you do df.loc[df['mynumbers']] you’re doing none of those 2 options. It’s trying to find the object df['mynumbers'] in the index, as shown by the following error, and that doesn’t work:

KeyError: "None of [Int64Index([51, 52, 53, 54, 55], dtype='int64')] are in the [index]"