sort index using lambda function with string split-CodePudding

I have a datafrmae lke this

df_crossplot 

the index is 1A22, 10A22,11A22,2A22,21A22
        value
1A22    10
10A22   12
11A22   11
2A22    15
12A22    21
3A22    25

What I like to do is sort index based on the number before A, like this

        value
1A22    10
2A22    15
3A22    25
10A22   12
11A22   11
12A22    21

The one I do is this with an error


df_crossplot=df_crossplot.sort_index(key=lambda x: float(x.str.split('A')[0]))


TypeError: float() argument must be a string or a number, not 'list'

seem like x inside lambda function is a list instead of each individual component of a series,

How to do it? Thanks

CodePudding user response：

use pandas func in key

df.sort_index(key=lambda x: x.str.split('A').str[0].astype('int'))

CodePudding user response：

Why not create a new column and sort by it?:

df['sort_col'] = df.index.str.split('A')[0].astype(int)

CodePudding user response：

natsort would do work easily if you want to sort index

from natsort import natsorted
print(df.reindex(index=natsorted(df.index)))