fda_colname='DBA Name_update',
sos_colname='Name',
id_var='IMPAQ_ID')
# turns out address fails to predict anything
# # try out all the different address columns for a match by address
# addr_out = mf.address_match_scoring(fda_row, temp_sos, update=True,
# addr_split=False,full_addr='Agent Address',id_var='IMPAQ_ID')
# addr_dict={}
# for j in addr_cols:
# addr_dict[j] = mf.address_match_scoring(fda_row, temp_sos, update=True,
# addr_split=False,full_addr=j,id_var='IMPAQ_ID')
#take max values
# for j in addr_cols:
# addr_out = pd.concat([addr_out, addr_dict[j]]).max(level=0)
# out = out.merge(addr_out, left_index=True, right_index=True, on='IMPAQ_ID')
name_match = mf.match_selection(out, 'basic name', name_lvl=.99)
# addr_match = mf.match_selection(out, 'basic addr',addrnum_lvl=.01,
# strname_lvl=.01, strtype_lvl=.01, zip_lvl=.01)
# if addr_match.empty == False:
# match = temp_sos.loc[addr_match.index,:]
if name_match.empty == False:
match = temp_sos.loc[name_match.index, :]
# if more than one tied, take the active record
if any(match['Status'].apply(
lambda x: x in ['Active', 'Good Standing'])):
match = match.loc[(match['Status'] == 'Active') |
(match['Status'] == 'Good Standing'), :]
name_match = name_match.loc[match.index, :]
match = match.loc[name_match.name_score.idxmax(), :]
# handling for corporation record
# filter SoS data to the same zip or town as the fda row to improve runtime
fda_row = fda_df.iloc[i, :]
temp_sos = sos_df.loc[(sos_df['pzip'] == fda_row['Zip_update']) |
(sos_df['pcity'] == fda_row['City_update']), :]
temp_sos.reset_index(inplace=True, drop=True)
# run through each matching algorithm
out = mf.name_match_scoring(fda_row,
temp_sos,
fda_colname='DBA Name_update',
sos_colname='bizname')
out = out.merge(mf.address_match_scoring(fda_row, temp_sos, update=True),
left_index=True,
right_index=True,
on='id')
out = temp_sos.merge(out, left_index=True, right_index=True, on='id')
match = mf.match_selection(out, 'basic addr')
match2 = mf.match_selection(out, 'basic name', name_lvl=.99)
matches.append([fda_row, match])
matches2.append([fda_row, match2])
if match.empty == False:
if isinstance(match, pd.DataFrame):
match = match.iloc[0, :]
fda_df.loc[i, 'SoS_record'] = match.id
fda_df.loc[i, 'entity'] = match.bizname
fda_df.loc[i, 'agent_name'] = match['Registered Agent Name']
fda_df.loc[i, 'agent_address'] = match['Registered Agent Address 1']
fda_df.loc[i, 'agent_city'] = match['Registered Agent City']
fda_df.loc[i, 'agent_state'] = match['Registered Agent State']
fda_df.loc[i, 'agent_zip'] = match['Registered Agent Zip']
elif match2.empty == False:
if isinstance(match2, pd.DataFrame):