def get_common_dataframe(variables, year=2006): """ Compare variables in erf an openfisca """ simulation = SurveySimulation() simulation.set_config(year=year) simulation.set_param() simulation.set_survey() simulation.compute() erf = ErfsDataTable(year=2006) if "ident" not in variables: erf_variables = variables + ["ident"] else: erf_variables = variables if "wprm" not in erf_variables: erf_variables = erf_variables + ["wprm"] else: erf_variables = erf_variables erf_dataframe = erf.get_values(erf_variables, table="menage") erf_dataframe.rename(columns={'ident': 'idmen'}, inplace=True) for col in erf_dataframe.columns: if col is not "idmen": erf_dataframe.rename(columns={col: col + "_erf"}, inplace=True) of_dataframe, of_dataframe_default = simulation.aggregated_by_entity( "men", variables, all_output_vars=False, force_sum=True) del of_dataframe_default merged_df = of_dataframe.merge(erf_dataframe, on="idmen") del of_dataframe, erf_dataframe return merged_df
def get_common_dataframe(variables, year = 2006): """ Compare variables in erf an openfisca """ simulation = SurveySimulation() simulation.set_config(year = year) simulation.set_param() simulation.set_survey() simulation.compute() erf = ErfsDataTable(year=2006) if "ident" not in variables: erf_variables = variables + ["ident"] else: erf_variables = variables if "wprm" not in erf_variables: erf_variables = erf_variables + ["wprm"] else: erf_variables = erf_variables erf_dataframe = erf.get_values(erf_variables, table="menage") erf_dataframe.rename(columns={'ident': 'idmen'}, inplace=True) for col in erf_dataframe.columns: if col is not "idmen": erf_dataframe.rename(columns={col: col + "_erf"}, inplace=True) of_dataframe, of_dataframe_default = simulation.aggregated_by_entity("men", variables, all_output_vars=False, force_sum=True) del of_dataframe_default merged_df = of_dataframe.merge(erf_dataframe, on="idmen") del of_dataframe, erf_dataframe return merged_df
def build_aggregates(): # writer = None years = range(2009, 2010) for year in years: yr = str(year) # fname = "Agg_%s.%s" %(str(yr), "xls") simu = SurveySimulation() simu.set_config(year=yr) simu.set_param() simu.set_survey() simu.compute() variables = ["garext", "ci_garext", "inthab", "ppe_brute", "rni"] x = simu.aggregated_by_entity("men", variables, all_output_vars=False) # df = x[0] # print df["ci_garext"].describe() agg = Aggregates() agg.set_simulation(simu) agg.show_default = False agg.show_real = False agg.show_diff = False agg.varlist = var_list # ERROR: var_list is undefined. agg.compute() cols = agg.aggr_frame.columns[:4] print agg.aggr_frame[cols].to_string() # if writer is None: # writer = ExcelWriter(str(fname_all)) # agg.aggr_frame.to_excel(writer, yr, index= False, header= True) del simu del agg import gc gc.collect()
def build_aggregates(): # writer = None years = range(2009,2010) for year in years: yr = str(year) # fname = "Agg_%s.%s" %(str(yr), "xls") simu = SurveySimulation() simu.set_config(year = yr) simu.set_param() simu.set_survey() simu.compute() variables = ["garext", "ci_garext", "inthab", "ppe_brute", "rni"] x = simu.aggregated_by_entity("men", variables, all_output_vars = False) # df = x[0] # print df["ci_garext"].describe() agg = Aggregates() agg.set_simulation(simu) agg.show_default = False agg.show_real = False agg.show_diff = False agg.varlist = var_list # ERROR: var_list is undefined. agg.compute() cols = agg.aggr_frame.columns[:4] print agg.aggr_frame[cols].to_string() # if writer is None: # writer = ExcelWriter(str(fname_all)) # agg.aggr_frame.to_excel(writer, yr, index= False, header= True) del simu del agg import gc gc.collect()