| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
pandas.dataframe.to_csv() by default exports data index as an column with empty column name. This causes uploading the export csv file to fail because AutoML Tables does not allow empty column names. Given that the data index is not useful for training the model. This PR fixes the problem by setting the index argument to false so that the index is not exported.
There was a problem hiding this comment.
LGTM
It appears the index=False option for to_csv has been in pandas for a long time.
P.S. What's the reason we require 0.24 and above?
google-cloud-python/automl/setup.py
Line 29 in 5cbfa26
Sorry, something went wrong.
|
Thanks, @tswast . I am not sure as I just become the maintainer for AutoML Tables' Python SDK. |
Sorry, something went wrong.
|
If you want to be consistent with BigQuery, we use 0.17.1. google-cloud-python/bigquery/setup.py Line 44 in cab728b Unless you count dropping Python 2 support in 0.25, pandas 0.17 was the version with the most breaking changes in recent history, so something after that will get most customers. |
Sorry, something went wrong.
Sorry, something went wrong.
pandas.dataframe.to_csv() by default exports data index as an column with empty column name. This causes uploading the export csv file to fail because AutoML Tables does not allow empty column names. Given that the data index is not useful for training the model. This PR fixes the problem by setting the index argument to false so that the index is not exported.
| Back | FazBrowse Home | New Git URL |
pandas.dataframe.to_csv() by default exports data index as an column
with empty column name. This causes uploading the export csv file to
fail because AutoML Tables does not allow empty column names. Given
that the data index is not useful for training the model. This PR
fixes the problem by setting the index argument to false so that the
index is not exported.
Fixes #9483