String Into Integer while sorting-CodePudding

Curious if there is a way to convert a string into an integer, only during the sort_values() process, or if it's easier to convert the variable to an integer prior to sorting and then convert back to string after sorting.

Current code ran, but code is not correct, because I believe the D_Index is actually a string, so it was sorted as 11,12,2,21,22,3 instead of 2,3,11,12,21,22. See table example and code below.

Model	D_Index
First	11
Second	12
Third	2
Fourth	21
Fifth	22
Sixth	3

df_New = df_Old.sort_values(['Model','D_Index'])

CodePudding user response：

You can pass a sorting key to sort_values:

out = df.sort_values(by='D_Index', key=lambda x: x.astype(int))

Output:

    Model D_Index
2   Third       2
5   Sixth       3
0   First      11
1  Second      12
3  Fourth      21
4   Fifth      22

If you want to include Model into it, you can use the Series' name:

df.sort_values(by=['Model','D_Index'], key=lambda x: x.astype(int) if x.name=='D_Index' else x)

Output (in your example, the sorting is trivial, since no Model has different D_Indexes):

    Model  D_Index
4   Fifth       22
0   First       11
3  Fourth       21
1  Second       12
5   Sixth        3
2   Third        2

CodePudding user response：

You can use a lambda function:

df.sort_values(by='D_Index', key=lambda col: col.astype(float))