CODEWITHSUNDEEP
pandas
Ralph
Ralph

Reputation: 161

Pandas dataframe to_csv() converts string "1" to "1.0" by default

Pandas dataframe to_csv() performs unwanted format conversion by default.

I have an Excel sheet that contains columns with integer values that refer to option items in a web application. I go to considerable lengths to convert these integer values to strings before exporting them to CSV. However, when I do, the to_csv() still converts a "1" to "1.0" and "0" to "0.0". First question is, how do I prevent Pandas from doing that? And second question, why does to_excel() NOT show this behavior. If I export my dataframe to Excel the integer values are correctly preserved.

I'm using Pandas 0.23.4 in a virtual environment on Mac OSX Mojave.

Any help is appreciated!

Upvotes: 1

Views: 1649

Answers (1)

EdChum
EdChum

Reputation: 394269

If you find that your types have changed to a float when originally you thought that it was integer or all strings, then it's most likely that you have missing values such as NaN, None.

So you can either drop those rows using dropna:

df[column] = df[column].dropna()

or replace/fill these values:

df[column] = df[column].fillna(some_val)

where some_val is 0, or whatever value you want.

This will then ensure the dtype for that column is not mixed or upcasted to float.

Upvotes: 1

Related Questions