python - How to sort numerically but read_csv with dtype=object? -

given simplified test.csv file:

wrong 8 7 6 3 1 2 4 5 9 10 

and code:

#!/usr/bin/python import pandas pd  data = pd.read_csv('test.csv', dtype=object) counts=data['wrong'].value_counts(dropna=false) counts_converted=counts.convert_objects(convert_numeric=true)  print counts_converted.sort_index() 

produces following output:

1     1 10    1 2     1 3     1 4     1 5     1 6     1 7     1 8     1 9     1 dtype: int64 

why last print statement not sort index 1-10?

i have force dtype object when csv file read overcome issues detecting mixed character, date, , numeric formats in columns, removing statement isn't going work me.

i thought convert series numeric, doesn't seem work.

editing question since commenting not allowing me use enter key without posting comment... [ahh, found many long rants feature. shift-enter works.]

@edchum suggested solution works simplified case, not work production data. consider less simple data file:

wrong,right 8,a 7,b 6,c 3,d 1, 2,f 4,g 5,h 9,i 10,j ,k 11,l 

the empty value on second last line causes error "cannot convert float nan integer."

i have many nans (all empty) need kept , counted in value_counts.

other empty cells seem turn large negative numbers (i.e. -5226413792388707240) upon casting int64.

apologies in advance obtuseness on part! help.

adding astype after reading makes sort properly.

you mention have sort out mixed characters , dates , stuff, before astype , should fine.

import pandas pd  data = pd.read_csv('/home/mikael/test.csv', dtype=object)  # sanitize data here  data['wrong'] = data['wrong'].astype(int) counts=data['wrong'].value_counts(dropna=false) counts_converted=counts.convert_objects(convert_numeric=true)  print counts_converted.sort_index()  1     1 2     1 3     1 4     1 5     1 6     1 7     1 8     1 9     1 10    1 dtype: int64