Pandas groupby and get most_frequent

pandas_groupby_mostfrequent.py

# Default way of handling groupby and mostfrequent is slow
df.groupby(groupby_column)[null_column].agg(lambda x: x.iat[0])

# if there are nan's in df:
df.groupby(groupby_column)[null_column].agg(lambda x: x.iat[0] if not x.isnull().all() else np.nan)

# faster way: use value_counts and keep first value
(df
 .groupby(groupby_column)[null_column]
 .value_counts(sort=True, dropna=False)
 .reset_index(name='Counts')
 .drop_duplicates(subset=groupby_column, keep='first')
 .set_index(groupby_column)
 .drop(columns='Counts')
 .squeeze()
)