This following command is saying that my categorical data cannot be >1-dimensional.
My categorical variables here are : law_lag (effective year of a law in a few states of the USA ) , year and statefip ( which is 20 years from 1990-2010 across 50 states [statefip ] of the United States ).
Can anyone tell me if there are any basic issues I'm missing here to address ? At a beginner level , there's a high chance I'll miss the basic stuff.
These are the info for my variables btw from my Dataframe after putting the info command:
year 93585 non-null object
statefip 93585 non-null category
law_lag 92924 non-null float32
My regression command - which is returning me the main error mentioned in the header- is the following one :
PH = PH[~pd.isnull(PH.ln_incwage)]
PH_male = PH[PH.male==1] formula = ( "ln_incwage ~ C(law_lag)*C(year) + C(statefip) + hispanic + asian + ismarried + lths + hsdegree + somecollege + age2" )
reg = ( smf .wls(formula, data=PH_male, weights=PH.black.values) .fit( cov_type='cluster',
cov_kwds={'groups': PH_male.fip.values}, method='pinv') )
reg.summary()