I have a database with two columns: name (str) and probability (float).

I am running this command:

df[['name','probability']].groupby('name').prod()

on a Databricks (runtime 7.3) notebook and df is a pyspark.pandas dataframe.

The error I get is:

PandasNotImplementedError: The method `pd.groupby.GroupBy.prod()` is not implemented yet.

I wonder if there is a workaround.

CodePudding user response：

In this case, I think your mistake is simply the fact that you have not the latest versions of pandas installed. From what I can see V.1.5.2, has such function in its documentation, and when I tried to run such group by on a sample data I succeeded. Try running this command on your shell that oughta upgrade your pandas version and you are gonna able to run such function.

pip install --upgrade pandas

CodePudding user response：

check with this type(df[['name','probability']].groupby('name')) is the type is pandas.core.groupby.generic.DataFrameGroupBy or else you want to update version