Pandas datafram fillna setting value based on datatype-CodePudding

I have a pandas dataframe that has data extract from a table. The table can be any given table from the database. The table may have unknown number of columns and datatypes. The database I am using is Redshift.

After get the data in pandas DF, I need to check for nan values in numeric/decimal type of columns if there are any columns with nan value, I need to change the value to 0

col_list_nan = pd.columns[pdf.isna().any()].tolist()

for i in col_list_nan:
  pdf = pdf[i].replace(np.nan, 0)

How can I change the column value and return complete dataframe with remaining unchanged columns and with the replaced columns that have 0 instead of nan

CodePudding user response：

Suppose you had data:

df = pd.DataFrame({'a': [0.5, np.nan, 1], 'b': ['string', 'ok', np.nan]})

that looks like:

     a       b
0  0.5  string
1  NaN      ok
2  1.0     NaN

Then you can select subset of columns using select_dtypes and replace NaN values with 0 using fillna:

s = df.select_dtypes(include=['float'])
df[s.columns] = s.fillna(0)

Output:

     a       b
0  0.5  string
1  0.0      ok
2  1.0     NaN