python - Pandas - custom rolling concatenation function per each group -

i have dataframe 3 columns

1. crop year
2. base_field_name
3. meta_crop

i want per each base_field_name per each of rows cumulative (rolling) concatenation of last 3 rows in meta_crop column.

trying use groupby, apply , rolling function cant make work relevant snippet.

def get_name_from_list(str_list):     print(str_list)     if len(str_list) == 0:         return ''      else:         return '_'.join([str(y) y in str_list if y])  def get_last_x_years_planting(x_years, data_ser):     ## data ser [(year, crop)] - assuming sorted ascending.     return data_ser.rolling(window=x_years, on='meta_crop').apply(lambda x: get_name_from_list(x))  work_df.sort_values(by='crop_year', ascending=true).groupby('base_field_name').apply(lambda x: get_last_x_years_planting(3, x)) 

i "typeerror - float required" error.