Python: Storing estimated coefficients from different regressions in a single matrix

Question

I have the below code which goes through my dataframe (df_main) and runs one regression (applying sm.OLS()) per each unique value of project_