CODEWITHSUNDEEP
pythonpandas
samuelbrody1249
samuelbrody1249

Reputation: 4767

Chaining three pandas statements

Is it possible to chain the following three statements together into one?

raw_df = df.copy().astype({'ids': 'str'})
raw_df['dels'] = np.where(raw_df['dels'] == 'nan', None, raw_df['dels'])
raw_df['langs'] = np.where(raw_df['langs'] == 'nan', None, raw_df['langs'])

If so, how could that be done?

Upvotes: 1

Views: 96

Answers (1)

Cameron Riddell
Cameron Riddell

Reputation: 13417

You can use the assign method to create new columns/overwrite existing columns. You can also shorten the logic here from np.where via Series.mask (or Series.replace) since all you want to do is swap our 'nan' for None and keep the rest of the column untouched.

Without seeing a copy of your data, I can't be certain this will work but you can try:

new_df = (raw_df.copy()
          .astype({"ids": "str"})
          .assign(
              dels=lambda df: df["dels"].mask(df["dels"] == 'nan', None),
              langs=lambda df: df["langs"].mask(df["langs"] == 'nan', None))
         )

Alternatively, if you just want to go ahead and replace 'nan' for None throughout the whole dataframe (and not just 2 columns) you can do it like this:

new_df = (raw_df.copy()
          .astype({"ids": "str"})
          .replace("nan", None))

Upvotes: 4

Related Questions