def disaggregate_pastureland(fba_w_sector, attr): """ The USDA CoA Cropland irrigated pastureland data only links to the 3 digit NAICS '112'. This function uses state level CoA 'Land in Farms' to allocate the county level acreage data to 6 digit NAICS. :param fba_w_sector: The CoA Cropland dataframe after linked to sectors :return: The CoA cropland dataframe with disaggregated pastureland data """ import flowsa from flowsa.flowbyfunctions import allocate_by_sector, clean_df, flow_by_activity_fields, \ fba_fill_na_dict # subset the coa data so only pastureland p = fba_w_sector.loc[fba_w_sector['Sector'] == '112'] # add temp loc column for state fips p.loc[:, 'Location_tmp'] = p['Location'].apply(lambda x: str(x[0:2])) # load usda coa cropland naics df_f = flowsa.getFlowByActivity(flowclass=['Land'], years=[attr['allocation_source_year']], datasource='USDA_CoA_Cropland_NAICS') df_f = clean_df(df_f, flow_by_activity_fields, fba_fill_na_dict) # subset to land in farms data df_f = df_f[df_f['FlowName'] == 'FARM OPERATIONS'] # subset to rows related to pastureland df_f = df_f.loc[df_f['ActivityConsumedBy'].apply(lambda x: str(x[0:3])) == '112'] # drop rows with "&' df_f = df_f[~df_f['ActivityConsumedBy'].str.contains('&')] # create sector column df_f.loc[:, 'Sector'] = df_f['ActivityConsumedBy'] # create proportional ratios df_f = allocate_by_sector(df_f, 'proportional') # drop naics = '11 df_f = df_f[df_f['Sector'] != '11'] # drop 000 in location df_f.loc[:, 'Location'] = df_f['Location'].apply(lambda x: str(x[0:2])) # merge the coa pastureland data with land in farm data df = p.merge(df_f[['Sector', 'Location', 'FlowAmountRatio']], how='left', left_on="Location_tmp", right_on="Location") # multiply the flowamount by the flowratio df.loc[:, 'FlowAmount'] = df['FlowAmount'] * df['FlowAmountRatio'] # drop columns and rename df = df.drop( columns=['Location_tmp', 'Sector_x', 'Location_y', 'FlowAmountRatio']) df = df.rename(columns={"Sector_y": "Sector", "Location_x": 'Location'}) # drop rows where sector = 112 and then concat with original fba_w_sector fba_w_sector = fba_w_sector[fba_w_sector['Sector'].apply( lambda x: str(x[0:3])) != '112'].reset_index(drop=True) fba_w_sector = pd.concat([fba_w_sector, df], sort=False).reset_index(drop=True) return fba_w_sector
def main(method_name): """ Creates a flowbysector dataset :param method_name: Name of method corresponding to flowbysector method yaml name :return: flowbysector """ log.info("Initiating flowbysector creation for " + method_name) # call on method method = load_method(method_name) # create dictionary of data and allocation datasets fb = method['source_names'] # Create empty list for storing fbs files fbs_list = [] for k, v in fb.items(): # pull fba data for allocation flows = load_source_dataframe(k, v) if v['data_format'] == 'FBA': # ensure correct datatypes and that all fields exist flows = clean_df(flows, flow_by_activity_fields, fba_fill_na_dict, drop_description=False) # clean up fba, if specified in yaml if v["clean_fba_df_fxn"] != 'None': log.info("Cleaning up " + k + " FlowByActivity") flows = getattr(sys.modules[__name__], v["clean_fba_df_fxn"])(flows) # if activity_sets are specified in a file, call them here if 'activity_set_file' in v: aset_names = pd.read_csv(flowbysectoractivitysetspath + v['activity_set_file'], dtype=str) # create dictionary of allocation datasets for different activities activities = v['activity_sets'] # subset activity data and allocate to sector for aset, attr in activities.items(): # subset by named activities if 'activity_set_file' in v: names = aset_names[aset_names['activity_set'] == aset]['name'] else: names = attr['names'] log.info("Preparing to handle subset of flownames " + ', '.join(map(str, names)) + " in " + k) # subset fba data by activity flows_subset = flows[ (flows[fba_activity_fields[0]].isin(names)) | (flows[fba_activity_fields[1]].isin(names))].reset_index( drop=True) # extract relevant geoscale data or aggregate existing data log.info("Subsetting/aggregating dataframe to " + attr['allocation_from_scale'] + " geoscale") flows_subset_geo = subset_df_by_geoscale( flows_subset, v['geoscale_to_use'], attr['allocation_from_scale']) # Add sectors to df activity, depending on level of specified sector aggregation log.info("Adding sectors to " + k) flow_subset_wsec = add_sectors_to_flowbyactivity( flows_subset_geo, sectorsourcename=method['target_sector_source'], allocationmethod=attr['allocation_method']) # clean up fba with sectors, if specified in yaml if v["clean_fba_w_sec_df_fxn"] != 'None': log.info("Cleaning up " + k + " FlowByActivity with sectors") flow_subset_wsec = getattr(sys.modules[__name__], v["clean_fba_w_sec_df_fxn"])( flow_subset_wsec, attr=attr) # map df to elementary flows log.info("Mapping flows in " + k + ' to federal elementary flow list') if 'fedefl_mapping' in v: mapping_files = v['fedefl_mapping'] else: mapping_files = k flow_subset_mapped = map_elementary_flows( flow_subset_wsec, mapping_files) # clean up mapped fba with sectors, if specified in yaml if "clean_mapped_fba_w_sec_df_fxn" in v: log.info("Cleaning up " + k + " FlowByActivity with sectors") flow_subset_mapped = getattr( sys.modules[__name__], v["clean_mapped_fba_w_sec_df_fxn"])(flow_subset_mapped, attr, method) # if allocation method is "direct", then no need to create alloc ratios, else need to use allocation # dataframe to create sector allocation ratios if attr['allocation_method'] == 'direct': log.info('Directly assigning ' + ', '.join(map(str, names)) + ' to sectors') fbs = flow_subset_mapped.copy() # for each activity, if activities are not sector like, check that there is no data loss if load_source_catalog( )[k]['sector-like_activities'] is False: activity_list = [] for n in names: log.info('Checking for ' + n + ' at ' + method['target_sector_level']) fbs_subset = fbs[( (fbs[fba_activity_fields[0]] == n) & (fbs[fba_activity_fields[1]] == n)) | (fbs[fba_activity_fields[0]] == n) | (fbs[fba_activity_fields[1]] == n )].reset_index(drop=True) fbs_subset = check_if_losing_sector_data( fbs_subset, method['target_sector_level']) activity_list.append(fbs_subset) fbs = pd.concat(activity_list, ignore_index=True) # if allocation method for an activity set requires a specific function due to the complicated nature # of the allocation, call on function here elif attr['allocation_method'] == 'allocation_function': log.info( 'Calling on function specified in method yaml to allocate ' + ', '.join(map(str, names)) + ' to sectors') fbs = getattr(sys.modules[__name__], attr['allocation_source'])( flow_subset_mapped, attr, fbs_list) else: # determine appropriate allocation dataset log.info("Loading allocation flowbyactivity " + attr['allocation_source'] + " for year " + str(attr['allocation_source_year'])) fba_allocation = flowsa.getFlowByActivity( flowclass=[attr['allocation_source_class']], datasource=attr['allocation_source'], years=[attr['allocation_source_year'] ]).reset_index(drop=True) # clean df and harmonize unites fba_allocation = clean_df(fba_allocation, flow_by_activity_fields, fba_fill_na_dict) fba_allocation = harmonize_units(fba_allocation) # check if allocation data exists at specified geoscale to use log.info("Checking if allocation data exists at the " + attr['allocation_from_scale'] + " level") check_if_data_exists_at_geoscale( fba_allocation, attr['allocation_from_scale']) # aggregate geographically to the scale of the flowbyactivty source, if necessary fba_allocation = subset_df_by_geoscale( fba_allocation, attr['allocation_from_scale'], v['geoscale_to_use']) # subset based on yaml settings if attr['allocation_flow'] != 'None': fba_allocation = fba_allocation.loc[ fba_allocation['FlowName'].isin( attr['allocation_flow'])] if attr['allocation_compartment'] != 'None': fba_allocation = fba_allocation.loc[ fba_allocation['Compartment'].isin( attr['allocation_compartment'])] # cleanup the fba allocation df, if necessary if 'clean_allocation_fba' in attr: log.info("Cleaning " + attr['allocation_source']) fba_allocation = getattr(sys.modules[__name__], attr["clean_allocation_fba"])( fba_allocation, attr=attr) # reset index fba_allocation = fba_allocation.reset_index(drop=True) # assign sector to allocation dataset log.info("Adding sectors to " + attr['allocation_source']) fba_allocation_wsec = add_sectors_to_flowbyactivity( fba_allocation, sectorsourcename=method['target_sector_source']) # call on fxn to further clean up/disaggregate the fba allocation data, if exists if 'clean_allocation_fba_w_sec' in attr: log.info("Further disaggregating sectors in " + attr['allocation_source']) fba_allocation_wsec = getattr( sys.modules[__name__], attr["clean_allocation_fba_w_sec"])( fba_allocation_wsec, attr=attr, method=method) # subset fba datasets to only keep the sectors associated with activity subset log.info("Subsetting " + attr['allocation_source'] + " for sectors in " + k) fba_allocation_subset = get_fba_allocation_subset( fba_allocation_wsec, k, names, flowSubsetMapped=flow_subset_mapped, allocMethod=attr['allocation_method']) # if there is an allocation helper dataset, modify allocation df if attr['allocation_helper'] == 'yes': log.info( "Using the specified allocation help for subset of " + attr['allocation_source']) fba_allocation_subset = allocation_helper( fba_allocation_subset, attr, method, v) # create flow allocation ratios for each activity # if load_source_catalog()[k]['sector-like_activities'] flow_alloc_list = [] group_cols = fba_mapped_default_grouping_fields group_cols = [ e for e in group_cols if e not in ('ActivityProducedBy', 'ActivityConsumedBy') ] for n in names: log.info("Creating allocation ratios for " + n) fba_allocation_subset_2 = get_fba_allocation_subset( fba_allocation_subset, k, [n], flowSubsetMapped=flow_subset_mapped, allocMethod=attr['allocation_method']) if len(fba_allocation_subset_2) == 0: log.info("No data found to allocate " + n) else: flow_alloc = allocate_by_sector( fba_allocation_subset_2, k, attr['allocation_source'], attr['allocation_method'], group_cols, flowSubsetMapped=flow_subset_mapped) flow_alloc = flow_alloc.assign(FBA_Activity=n) flow_alloc_list.append(flow_alloc) flow_allocation = pd.concat(flow_alloc_list, ignore_index=True) # generalize activity field names to enable link to main fba source log.info("Generalizing activity columns in subset of " + attr['allocation_source']) flow_allocation = collapse_activity_fields(flow_allocation) # check for issues with allocation ratios check_allocation_ratios(flow_allocation, aset, k, method_name) # create list of sectors in the flow allocation df, drop any rows of data in the flow df that \ # aren't in list sector_list = flow_allocation['Sector'].unique().tolist() # subset fba allocation table to the values in the activity list, based on overlapping sectors flow_subset_mapped = flow_subset_mapped.loc[ (flow_subset_mapped[fbs_activity_fields[0]]. isin(sector_list)) | (flow_subset_mapped[fbs_activity_fields[1]]. isin(sector_list))] # check if fba and allocation dfs have the same LocationSystem log.info( "Checking if flowbyactivity and allocation dataframes use the same location systems" ) check_if_location_systems_match(flow_subset_mapped, flow_allocation) # merge fba df w/flow allocation dataset log.info("Merge " + k + " and subset of " + attr['allocation_source']) fbs = flow_subset_mapped.merge( flow_allocation[[ 'Location', 'Sector', 'FlowAmountRatio', 'FBA_Activity' ]], left_on=[ 'Location', 'SectorProducedBy', 'ActivityProducedBy' ], right_on=['Location', 'Sector', 'FBA_Activity'], how='left') fbs = fbs.merge( flow_allocation[[ 'Location', 'Sector', 'FlowAmountRatio', 'FBA_Activity' ]], left_on=[ 'Location', 'SectorConsumedBy', 'ActivityConsumedBy' ], right_on=['Location', 'Sector', 'FBA_Activity'], how='left') # merge the flowamount columns fbs.loc[:, 'FlowAmountRatio'] = fbs[ 'FlowAmountRatio_x'].fillna(fbs['FlowAmountRatio_y']) # fill null rows with 0 because no allocation info fbs['FlowAmountRatio'] = fbs['FlowAmountRatio'].fillna(0) # check if fba and alloc dfs have data for same geoscales - comment back in after address the 'todo' # log.info("Checking if flowbyactivity and allocation dataframes have data at the same locations") # check_if_data_exists_for_same_geoscales(fbs, k, attr['names']) # drop rows where there is no allocation data fbs = fbs.dropna(subset=['Sector_x', 'Sector_y'], how='all').reset_index() # calculate flow amounts for each sector log.info("Calculating new flow amounts using flow ratios") fbs.loc[:, 'FlowAmount'] = fbs['FlowAmount'] * fbs[ 'FlowAmountRatio'] # drop columns log.info("Cleaning up new flow by sector") fbs = fbs.drop(columns=[ 'Sector_x', 'FlowAmountRatio_x', 'Sector_y', 'FlowAmountRatio_y', 'FlowAmountRatio', 'FBA_Activity_x', 'FBA_Activity_y' ]) # drop rows where flowamount = 0 (although this includes dropping suppressed data) fbs = fbs[fbs['FlowAmount'] != 0].reset_index(drop=True) # define grouping columns dependent on sectors being activity-like or not if load_source_catalog()[k]['sector-like_activities'] is False: groupingcols = fbs_grouping_fields_w_activities groupingdict = flow_by_sector_fields_w_activity else: groupingcols = fbs_default_grouping_fields groupingdict = flow_by_sector_fields # clean df fbs = clean_df(fbs, groupingdict, fbs_fill_na_dict) # aggregate df geographically, if necessary # todo: replace with fxn return_from_scale log.info("Aggregating flowbysector to " + method['target_geoscale'] + " level") if fips_number_key[v['geoscale_to_use']] < fips_number_key[ attr['allocation_from_scale']]: from_scale = v['geoscale_to_use'] else: from_scale = attr['allocation_from_scale'] to_scale = method['target_geoscale'] fbs_geo_agg = agg_by_geoscale(fbs, from_scale, to_scale, groupingcols) # aggregate data to every sector level log.info("Aggregating flowbysector to all sector levels") fbs_sec_agg = sector_aggregation(fbs_geo_agg, groupingcols) # add missing naics5/6 when only one naics5/6 associated with a naics4 fbs_agg = sector_disaggregation(fbs_sec_agg, groupingdict) # check if any sector information is lost before reaching the target sector length, if so, # allocate values equally to disaggregated sectors log.info('Checking for data at ' + method['target_sector_level']) fbs_agg_2 = check_if_losing_sector_data( fbs_agg, method['target_sector_level']) # compare flowbysector with flowbyactivity # todo: modify fxn to work if activities are sector like in df being allocated if load_source_catalog()[k]['sector-like_activities'] is False: check_for_differences_between_fba_load_and_fbs_output( flow_subset_mapped, fbs_agg_2, aset, k, method_name) # return sector level specified in method yaml # load the crosswalk linking sector lengths sector_list = get_sector_list(method['target_sector_level']) # subset df, necessary because not all of the sectors are NAICS and can get duplicate rows fbs_1 = fbs_agg_2.loc[ (fbs_agg_2[fbs_activity_fields[0]].isin(sector_list)) & (fbs_agg_2[fbs_activity_fields[1]].isin(sector_list) )].reset_index(drop=True) fbs_2 = fbs_agg_2.loc[ (fbs_agg_2[fbs_activity_fields[0]].isin(sector_list)) & (fbs_agg_2[fbs_activity_fields[1]].isnull())].reset_index( drop=True) fbs_3 = fbs_agg_2.loc[ (fbs_agg_2[fbs_activity_fields[0]].isnull()) & (fbs_agg_2[fbs_activity_fields[1]].isin(sector_list) )].reset_index(drop=True) fbs_sector_subset = pd.concat([fbs_1, fbs_2, fbs_3]) # drop activity columns fbs_sector_subset = fbs_sector_subset.drop( ['ActivityProducedBy', 'ActivityConsumedBy'], axis=1, errors='ignore') # save comparison of FBA total to FBS total for an activity set compare_fba_load_and_fbs_output_totals(flows_subset_geo, fbs_sector_subset, aset, k, method_name, attr, method, mapping_files) log.info( "Completed flowbysector for activity subset with flows " + ', '.join(map(str, names))) fbs_list.append(fbs_sector_subset) else: # if the loaded flow dt is already in FBS format, append directly to list of FBS log.info("Append " + k + " to FBS list") # ensure correct field datatypes and add any missing fields flows = clean_df(flows, flow_by_sector_fields, fbs_fill_na_dict) fbs_list.append(flows) # create single df of all activities log.info("Concat data for all activities") fbss = pd.concat(fbs_list, ignore_index=True, sort=False) log.info("Clean final dataframe") # aggregate df as activities might have data for the same specified sector length fbss = clean_df(fbss, flow_by_sector_fields, fbs_fill_na_dict) fbss = aggregator(fbss, fbs_default_grouping_fields) # sort df log.info("Sort and store dataframe") # add missing fields, ensure correct data type, reorder columns fbss = fbss.sort_values( ['SectorProducedBy', 'SectorConsumedBy', 'Flowable', 'Context']).reset_index(drop=True) # save parquet file store_flowbysector(fbss, method_name)
def disaggregate_pastureland(fba_w_sector, attr, method, years_list, sector_column): """ The USDA CoA Cropland irrigated pastureland data only links to the 3 digit NAICS '112'. This function uses state level CoA 'Land in Farms' to allocate the county level acreage data to 6 digit NAICS. :param fba_w_sector: The CoA Cropland dataframe after linked to sectors :param attr: :param years_list: :param sector_column: The sector column on which to make df modifications (SectorProducedBy or SectorConsumedBy) :return: The CoA cropland dataframe with disaggregated pastureland data """ import flowsa from flowsa.flowbyfunctions import allocate_by_sector, clean_df, flow_by_activity_fields, \ fba_fill_na_dict, replace_strings_with_NoneType, replace_NoneType_with_empty_cells, \ fba_mapped_default_grouping_fields, harmonize_units from flowsa.mapping import add_sectors_to_flowbyactivity # tmp drop NoneTypes fba_w_sector = replace_NoneType_with_empty_cells(fba_w_sector) # subset the coa data so only pastureland p = fba_w_sector.loc[fba_w_sector[sector_column].apply(lambda x: x[0:3]) == '112'].reset_index(drop=True) if len(p) != 0: # add temp loc column for state fips p = p.assign(Location_tmp=p['Location'].apply(lambda x: x[0:2])) df_sourcename = pd.unique(p['SourceName'])[0] # load usda coa cropland naics df_class = ['Land'] df_years = years_list df_allocation = 'USDA_CoA_Cropland_NAICS' df_f = flowsa.getFlowByActivity(flowclass=df_class, years=df_years, datasource=df_allocation) df_f = clean_df(df_f, flow_by_activity_fields, fba_fill_na_dict) df_f = harmonize_units(df_f) # subset to land in farms data df_f = df_f[df_f['FlowName'] == 'FARM OPERATIONS'] # subset to rows related to pastureland df_f = df_f.loc[df_f['ActivityConsumedBy'].apply(lambda x: x[0:3]) == '112'] # drop rows with "&' df_f = df_f[~df_f['ActivityConsumedBy'].str.contains('&')] # create sector columns df_f = add_sectors_to_flowbyactivity(df_f, sectorsourcename=method['target_sector_source']) # create proportional ratios group_cols = fba_mapped_default_grouping_fields group_cols = [e for e in group_cols if e not in ('ActivityProducedBy', 'ActivityConsumedBy')] df_f = allocate_by_sector(df_f, df_sourcename, df_allocation, 'proportional', group_cols) # tmp drop NoneTypes df_f = replace_NoneType_with_empty_cells(df_f) # drop naics = '11 df_f = df_f[df_f[sector_column] != '11'] # drop 000 in location df_f = df_f.assign(Location=df_f['Location'].apply(lambda x: x[0:2])) # merge the coa pastureland data with land in farm data df = p.merge(df_f[[sector_column, 'Location', 'FlowAmountRatio']], how='left', left_on="Location_tmp", right_on="Location") # multiply the flowamount by the flowratio df.loc[:, 'FlowAmount'] = df['FlowAmount'] * df['FlowAmountRatio'] # drop columns and rename df = df.drop(columns=['Location_tmp', sector_column + '_x', 'Location_y', 'FlowAmountRatio']) df = df.rename(columns={sector_column + '_y': sector_column, "Location_x": 'Location'}) # drop rows where sector = 112 and then concat with original fba_w_sector fba_w_sector = fba_w_sector[fba_w_sector[sector_column].apply(lambda x: x[0:3]) != '112'].reset_index(drop=True) fba_w_sector = pd.concat([fba_w_sector, df], sort=True).reset_index(drop=True) # fill empty cells with NoneType fba_w_sector = replace_strings_with_NoneType(fba_w_sector) return fba_w_sector
def main(method_name): """ Creates a flowbysector dataset :param method_name: Name of method corresponding to flowbysector method yaml name :return: flowbysector """ log.info("Initiating flowbysector creation for " + method_name) # call on method method = load_method(method_name) # create dictionary of data and allocation datasets fb = method['source_names'] # Create empty list for storing fbs files fbss = [] for k, v in fb.items(): # pull fba data for allocation flows = load_source_dataframe(k, v) if v['data_format'] == 'FBA': # clean up fba, if specified in yaml if v["clean_fba_df_fxn"] != 'None': log.info("Cleaning up " + k + " FlowByActivity") flows = getattr(sys.modules[__name__], v["clean_fba_df_fxn"])(flows) flows = clean_df(flows, flow_by_activity_fields, fba_fill_na_dict) # create dictionary of allocation datasets for different activities activities = v['activity_sets'] # subset activity data and allocate to sector for aset, attr in activities.items(): # subset by named activities names = attr['names'] log.info("Preparing to handle subset of flownames " + ', '.join(map(str, names)) + " in " + k) # check if flowbyactivity data exists at specified geoscale to use flow_subset_list = [] for n in names: # subset usgs data by activity flow_subset = flows[(flows[fba_activity_fields[0]] == n) | (flows[fba_activity_fields[1]] == n)].reset_index(drop=True) log.info("Checking if flowbyactivity data exists for " + n + " at the " + v['geoscale_to_use'] + ' level') geocheck = check_if_data_exists_at_geoscale(flow_subset, v['geoscale_to_use'], activitynames=n) # aggregate geographically to the scale of the allocation dataset if geocheck == "Yes": activity_from_scale = v['geoscale_to_use'] else: # if activity does not exist at specified geoscale, issue warning and use data at less aggregated # geoscale, and sum to specified geoscale log.info("Checking if flowbyactivity data exists for " + n + " at a less aggregated level") activity_from_scale = check_if_data_exists_at_less_aggregated_geoscale(flow_subset, v['geoscale_to_use'], n) activity_to_scale = attr['allocation_from_scale'] # if df is less aggregated than allocation df, aggregate usgs activity to allocation geoscale if fips_number_key[activity_from_scale] > fips_number_key[activity_to_scale]: log.info("Aggregating subset from " + activity_from_scale + " to " + activity_to_scale) flow_subset = agg_by_geoscale(flow_subset, activity_from_scale, activity_to_scale, fba_default_grouping_fields, n) # else, aggregate to geoscale want to use elif fips_number_key[activity_from_scale] > fips_number_key[v['geoscale_to_use']]: log.info("Aggregating subset from " + activity_from_scale + " to " + v['geoscale_to_use']) flow_subset = agg_by_geoscale(flow_subset, activity_from_scale, v['geoscale_to_use'], fba_default_grouping_fields, n) # else, if usgs is more aggregated than allocation table, filter relevant rows else: log.info("Subsetting " + activity_from_scale + " data") flow_subset = filter_by_geoscale(flow_subset, activity_from_scale, n) # Add sectors to df activity, depending on level of specified sector aggregation log.info("Adding sectors to " + k + " for " + n) flow_subset_wsec = add_sectors_to_flowbyactivity(flow_subset, sectorsourcename=method['target_sector_source'], levelofSectoragg=attr['activity_sector_aggregation']) flow_subset_list.append(flow_subset_wsec) flow_subset_wsec = pd.concat(flow_subset_list, sort=False).reset_index(drop=True) # clean up fba with sectors, if specified in yaml if v["clean_fba_w_sec_df_fxn"] != 'None': log.info("Cleaning up " + k + " FlowByActivity with sectors") flow_subset_wsec = getattr(sys.modules[__name__], v["clean_fba_w_sec_df_fxn"])(flow_subset_wsec, attr) # map df to elementary flows - commented out until mapping complete log.info("Mapping flows in " + k + ' to federal elementary flow list') flow_subset_wsec = map_elementary_flows(flow_subset_wsec, k) # if allocation method is "direct", then no need to create alloc ratios, else need to use allocation # dataframe to create sector allocation ratios if attr['allocation_method'] == 'direct': log.info('Directly assigning ' + ', '.join(map(str, names)) + ' to sectors') fbs = flow_subset_wsec.copy() else: # determine appropriate allocation dataset log.info("Loading allocation flowbyactivity " + attr['allocation_source'] + " for year " + str(attr['allocation_source_year'])) fba_allocation = flowsa.getFlowByActivity(flowclass=[attr['allocation_source_class']], datasource=attr['allocation_source'], years=[attr['allocation_source_year']]).reset_index(drop=True) fba_allocation = clean_df(fba_allocation, flow_by_activity_fields, fba_fill_na_dict) # subset based on yaml settings if attr['allocation_flow'] != 'None': fba_allocation = fba_allocation.loc[fba_allocation['FlowName'].isin(attr['allocation_flow'])] if attr['allocation_compartment'] != 'None': fba_allocation = fba_allocation.loc[ fba_allocation['Compartment'].isin(attr['allocation_compartment'])] # cleanup the fba allocation df, if necessary if 'clean_allocation_fba' in attr: log.info("Cleaning " + attr['allocation_source']) fba_allocation = getattr(sys.modules[__name__], attr["clean_allocation_fba"])(fba_allocation) # reset index fba_allocation = fba_allocation.reset_index(drop=True) # check if allocation data exists at specified geoscale to use log.info("Checking if allocation data exists at the " + attr['allocation_from_scale'] + " level") check_if_data_exists_at_geoscale(fba_allocation, attr['allocation_from_scale']) # aggregate geographically to the scale of the flowbyactivty source, if necessary from_scale = attr['allocation_from_scale'] to_scale = v['geoscale_to_use'] # if allocation df is less aggregated than FBA df, aggregate allocation df to target scale if fips_number_key[from_scale] > fips_number_key[to_scale]: fba_allocation = agg_by_geoscale(fba_allocation, from_scale, to_scale, fba_default_grouping_fields, names) # else, if usgs is more aggregated than allocation table, use usgs as both to and from scale else: fba_allocation = filter_by_geoscale(fba_allocation, from_scale, names) # assign sector to allocation dataset log.info("Adding sectors to " + attr['allocation_source']) fba_allocation = add_sectors_to_flowbyactivity(fba_allocation, sectorsourcename=method['target_sector_source'], levelofSectoragg=attr['allocation_sector_aggregation']) # subset fba datsets to only keep the sectors associated with activity subset log.info("Subsetting " + attr['allocation_source'] + " for sectors in " + k) fba_allocation_subset = get_fba_allocation_subset(fba_allocation, k, names) # generalize activity field names to enable link to main fba source log.info("Generalizing activity columns in subset of " + attr['allocation_source']) fba_allocation_subset = generalize_activity_field_names(fba_allocation_subset) # drop columns fba_allocation_subset = fba_allocation_subset.drop(columns=['Activity']) # call on fxn to further disaggregate the fba allocation data, if exists if 'allocation_disaggregation_fxn' in attr: log.info("Futher disaggregating sectors in " + attr['allocation_source']) fba_allocation_subset = getattr(sys.modules[__name__], attr["allocation_disaggregation_fxn"])(fba_allocation_subset, attr) # if there is an allocation helper dataset, modify allocation df if attr['allocation_helper'] == 'yes': log.info("Using the specified allocation help for subset of " + attr['allocation_source']) fba_allocation_subset = allocation_helper(fba_allocation_subset, method, attr) # create flow allocation ratios log.info("Creating allocation ratios for " + attr['allocation_source']) flow_allocation = allocate_by_sector(fba_allocation_subset, attr['allocation_method']) # create list of sectors in the flow allocation df, drop any rows of data in the flow df that \ # aren't in list sector_list = flow_allocation['Sector'].unique().tolist() # subset fba allocation table to the values in the activity list, based on overlapping sectors flow_subset_wsec = flow_subset_wsec.loc[ (flow_subset_wsec[fbs_activity_fields[0]].isin(sector_list)) | (flow_subset_wsec[fbs_activity_fields[1]].isin(sector_list))] # check if fba and allocation dfs have the same LocationSystem log.info("Checking if flowbyactivity and allocation dataframes use the same location systems") check_if_location_systems_match(flow_subset_wsec, flow_allocation) # merge fba df w/flow allocation dataset log.info("Merge " + k + " and subset of " + attr['allocation_source']) fbs = flow_subset_wsec.merge( flow_allocation[['Location', 'Sector', 'FlowAmountRatio']], left_on=['Location', 'SectorProducedBy'], right_on=['Location', 'Sector'], how='left') fbs = fbs.merge( flow_allocation[['Location', 'Sector', 'FlowAmountRatio']], left_on=['Location', 'SectorConsumedBy'], right_on=['Location', 'Sector'], how='left') # merge the flowamount columns fbs.loc[:, 'FlowAmountRatio'] = fbs['FlowAmountRatio_x'].fillna(fbs['FlowAmountRatio_y']) # check if fba and alloc dfs have data for same geoscales - comment back in after address the 'todo' # log.info("Checking if flowbyactivity and allocation dataframes have data at the same locations") # check_if_data_exists_for_same_geoscales(fbs, k, attr['names']) # drop rows where there is no allocation data fbs = fbs.dropna(subset=['Sector_x', 'Sector_y'], how='all').reset_index() # calculate flow amounts for each sector log.info("Calculating new flow amounts using flow ratios") fbs.loc[:, 'FlowAmount'] = fbs['FlowAmount'] * fbs['FlowAmountRatio'] # drop columns log.info("Cleaning up new flow by sector") fbs = fbs.drop(columns=['Sector_x', 'FlowAmountRatio_x', 'Sector_y', 'FlowAmountRatio_y', 'FlowAmountRatio', 'ActivityProducedBy', 'ActivityConsumedBy']) # drop rows where flowamount = 0 (although this includes dropping suppressed data) fbs = fbs[fbs['FlowAmount'] != 0].reset_index(drop=True) # clean df fbs = clean_df(fbs, flow_by_sector_fields, fbs_fill_na_dict) # aggregate df geographically, if necessary log.info("Aggregating flowbysector to " + method['target_geoscale'] + " level") if fips_number_key[v['geoscale_to_use']] < fips_number_key[attr['allocation_from_scale']]: from_scale = v['geoscale_to_use'] else: from_scale = attr['allocation_from_scale'] to_scale = method['target_geoscale'] fbs = agg_by_geoscale(fbs, from_scale, to_scale, fbs_default_grouping_fields, names) # aggregate data to every sector level log.info("Aggregating flowbysector to all sector levels") fbs = sector_aggregation(fbs, fbs_default_grouping_fields) # add missing naics5/6 when only one naics5/6 associated with a naics4 fbs = sector_disaggregation(fbs) # test agg by sector # sector_agg_comparison = sector_flow_comparision(fbs) # return sector level specified in method yaml # load the crosswalk linking sector lengths sector_list = get_sector_list(method['target_sector_level']) # add any non-NAICS sectors used with NAICS sector_list = add_non_naics_sectors(sector_list, method['target_sector_level']) # subset df, necessary because not all of the sectors are NAICS and can get duplicate rows fbs_1 = fbs.loc[(fbs[fbs_activity_fields[0]].isin(sector_list)) & (fbs[fbs_activity_fields[1]].isin(sector_list))].reset_index(drop=True) fbs_2 = fbs.loc[(fbs[fbs_activity_fields[0]].isin(sector_list)) | (fbs[fbs_activity_fields[1]].isin(sector_list))].reset_index(drop=True) fbs_sector_subset = pd.concat([fbs_1, fbs_2], sort=False) # set source name fbs_sector_subset.loc[:, 'SectorSourceName'] = method['target_sector_source'] log.info("Completed flowbysector for activity subset with flows " + ', '.join(map(str, names))) fbss.append(fbs_sector_subset) else: # if the loaded flow dt is already in FBS format, append directly to list of FBS log.info("Append " + k + " to FBS list") fbss.append(flows) # create single df of all activities log.info("Concat data for all activities") fbss = pd.concat(fbss, ignore_index=True, sort=False) log.info("Clean final dataframe") # aggregate df as activities might have data for the same specified sector length fbss = aggregator(fbss, fbs_default_grouping_fields) # sort df log.info("Sort and store dataframe") fbss = fbss.replace({'nan': None}) # add missing fields, ensure correct data type, reorder columns fbss = clean_df(fbss, flow_by_sector_fields, fbs_fill_na_dict) fbss = fbss.sort_values( ['SectorProducedBy', 'SectorConsumedBy', 'Flowable', 'Context']).reset_index(drop=True) # save parquet file store_flowbysector(fbss, method_name)
def main(method_name): """ Creates a flowbysector dataset :param method_name: Name of method corresponding to flowbysector method yaml name :return: flowbysector """ log.info("Initiating flowbysector creation for " + method_name) # call on method method = load_method(method_name) # create dictionary of water data and allocation datasets fbas = method['flowbyactivity_sources'] # Create empty list for storing fbs files fbss = [] for k, v in fbas.items(): # pull water data for allocation log.info("Retrieving flowbyactivity for datasource " + k + " in year " + str(v['year'])) flows = flowsa.getFlowByActivity(flowclass=[v['class']], years=[v['year']], datasource=k) # if necessary, standardize names in data set if v['activity_name_standardization_fxn'] != 'None': log.info("Standardizing activity names in " + k) flows = getattr(sys.modules[__name__], v['activity_name_standardization_fxn'])(flows) # drop description field flows = flows.drop(columns='Description') # fill null values flows = flows.fillna(value=fba_fill_na_dict) # map df to elementary flows - commented out until mapping complete # log.info("Mapping flows in " + k + ' to federal elementary flow list') # flows_mapped = map_elementary_flows(flows, k) # convert unit todo: think about unit conversion here log.info("Converting units in " + k) flows = convert_unit(flows) # create dictionary of allocation datasets for different activities activities = v['activity_sets'] for aset, attr in activities.items(): # subset by named activities names = [attr['names']] log.info("Preparing to handle subset of flownames " + ', '.join(map(str, names)) + " in " + k) # subset usgs data by activity flow_subset = flows[(flows[fba_activity_fields[0]].isin(names)) | (flows[fba_activity_fields[1]].isin(names))] # Reset index values after subset flow_subset = flow_subset.reset_index(drop=True) # check if flowbyactivity data exists at specified geoscale to use log.info("Checking if flowbyactivity data exists for " + ', '.join(map(str, names)) + " at the " + v['geoscale_to_use'] + ' level') geocheck = check_if_data_exists_at_geoscale(flow_subset, names, v['geoscale_to_use']) # aggregate geographically to the scale of the allocation dataset if geocheck == "Yes": activity_from_scale = v['geoscale_to_use'] else: # if activity does not exist at specified geoscale, issue warning and use data at less aggregated # geoscale, and sum to specified geoscale log.info("Checking if flowbyactivity data exists for " + ', '.join(map(str, names)) + " at a less aggregated level") new_geoscale_to_use = check_if_data_exists_at_less_aggregated_geoscale(flow_subset, names, v['geoscale_to_use']) activity_from_scale = new_geoscale_to_use activity_to_scale = attr['allocation_from_scale'] # if usgs is less aggregated than allocation df, aggregate usgs activity to target scale if fips_number_key[activity_from_scale] > fips_number_key[activity_to_scale]: log.info("Aggregating subset from " + activity_from_scale + " to " + activity_to_scale) flow_subset = agg_by_geoscale(flow_subset, activity_from_scale, activity_to_scale, fba_default_grouping_fields, names) # else, aggregate to geoscale want to use elif fips_number_key[activity_from_scale] > fips_number_key[v['geoscale_to_use']]: log.info("Aggregating subset from " + activity_from_scale + " to " + v['geoscale_to_use']) flow_subset = agg_by_geoscale(flow_subset, activity_from_scale, v['geoscale_to_use'], fba_default_grouping_fields, names) # else, if usgs is more aggregated than allocation table, filter relevant rows else: log.info("Filtering out " + activity_from_scale + " data") flow_subset = filter_by_geoscale(flow_subset, activity_from_scale, names) # location column pad zeros if necessary flow_subset['Location'] = flow_subset['Location'].apply(lambda x: x.ljust(3 + len(x), '0') if len(x) < 5 else x ) # Add sectors to usgs activity, creating two versions of the flow subset # the first version "flow_subset" is the most disaggregated version of the Sectors (NAICS) # the second version, "flow_subset_agg" includes only the most aggregated level of sectors log.info("Adding sectors to " + k + " for " + ', '.join(map(str, names))) flow_subset_wsec = add_sectors_to_flowbyactivity(flow_subset, sectorsourcename=method['target_sector_source']) flow_subset_wsec_agg = add_sectors_to_flowbyactivity(flow_subset, sectorsourcename=method['target_sector_source'], levelofSectoragg='agg') # if allocation method is "direct", then no need to create alloc ratios, else need to use allocation # dataframe to create sector allocation ratios if attr['allocation_method'] == 'direct': fbs = flow_subset_wsec_agg.copy() else: # determine appropriate allocation dataset log.info("Loading allocation flowbyactivity " + attr['allocation_source'] + " for year " + str(attr['allocation_source_year'])) fba_allocation = flowsa.getFlowByActivity(flowclass=[attr['allocation_source_class']], datasource=attr['allocation_source'], years=[attr['allocation_source_year']]).reset_index(drop=True) # fill null values fba_allocation = fba_allocation.fillna(value=fba_fill_na_dict) # convert unit fba_allocation = convert_unit(fba_allocation) # subset based on yaml settings if attr['allocation_flow'] != 'None': fba_allocation = fba_allocation.loc[fba_allocation['FlowName'].isin(attr['allocation_flow'])] if attr['allocation_compartment'] != 'None': fba_allocation = fba_allocation.loc[ fba_allocation['Compartment'].isin(attr['allocation_compartment'])] # reset index fba_allocation = fba_allocation.reset_index(drop=True) # check if allocation data exists at specified geoscale to use log.info("Checking if" + " allocation data exists for " + ', '.join(map(str, names)) + " at the " + attr['allocation_from_scale'] + " level") check_if_data_exists_at_geoscale(fba_allocation, names, attr['allocation_from_scale']) # aggregate geographically to the scale of the flowbyactivty source, if necessary from_scale = attr['allocation_from_scale'] to_scale = v['geoscale_to_use'] # if allocation df is less aggregated than FBA df, aggregate allocation df to target scale if fips_number_key[from_scale] > fips_number_key[to_scale]: fba_allocation = agg_by_geoscale(fba_allocation, from_scale, to_scale, fba_default_grouping_fields, names) # else, if usgs is more aggregated than allocation table, use usgs as both to and from scale else: fba_allocation = filter_by_geoscale(fba_allocation, from_scale, names) # assign sector to allocation dataset log.info("Adding sectors to " + attr['allocation_source']) fba_allocation = add_sectors_to_flowbyactivity(fba_allocation, sectorsourcename=method['target_sector_source'], levelofSectoragg=attr[ 'allocation_sector_aggregation']) # subset fba datsets to only keep the naics associated with usgs activity subset log.info("Subsetting " + attr['allocation_source'] + " for sectors in " + k) fba_allocation_subset = get_fba_allocation_subset(fba_allocation, k, names) # Reset index values after subset fba_allocation_subset = fba_allocation_subset.reset_index(drop=True) # generalize activity field names to enable link to water withdrawal table log.info("Generalizing activity names in subset of " + attr['allocation_source']) fba_allocation_subset = generalize_activity_field_names(fba_allocation_subset) # drop columns fba_allocation_subset = fba_allocation_subset.drop(columns=['Activity']) # if there is an allocation helper dataset, modify allocation df if attr['allocation_helper'] == 'yes': log.info("Using the specified allocation help for subset of " + attr['allocation_source']) fba_allocation_subset = allocation_helper(fba_allocation_subset, method, attr) # create flow allocation ratios log.info("Creating allocation ratios for " + attr['allocation_source']) flow_allocation = allocate_by_sector(fba_allocation_subset, attr['allocation_method']) # create list of sectors in the flow allocation df, drop any rows of data in the flow df that \ # aren't in list sector_list = flow_allocation['Sector'].unique().tolist() # subset fba allocation table to the values in the activity list, based on overlapping sectors flow_subset_wsec = flow_subset_wsec.loc[ (flow_subset_wsec[fbs_activity_fields[0]].isin(sector_list)) | (flow_subset_wsec[fbs_activity_fields[1]].isin(sector_list))] # check if fba and allocation dfs have the same LocationSystem log.info("Checking if flowbyactivity and allocation dataframes use the same location systems") check_if_location_systems_match(flow_subset_wsec, flow_allocation) # merge water withdrawal df w/flow allocation dataset log.info("Merge " + k + " and subset of " + attr['allocation_source']) fbs = flow_subset_wsec.merge( flow_allocation[['Location', 'LocationSystem', 'Sector', 'FlowAmountRatio']], left_on=['Location', 'LocationSystem', 'SectorProducedBy'], right_on=['Location', 'LocationSystem', 'Sector'], how='left') fbs = fbs.merge( flow_allocation[['Location', 'LocationSystem', 'Sector', 'FlowAmountRatio']], left_on=['Location', 'LocationSystem', 'SectorConsumedBy'], right_on=['Location', 'LocationSystem', 'Sector'], how='left') # drop columns where both sector produced/consumed by in flow allocation dif is null fbs = fbs.dropna(subset=['Sector_x', 'Sector_y'], how='all').reset_index() # merge the flowamount columns fbs['FlowAmountRatio'] = fbs['FlowAmountRatio_x'].fillna(fbs['FlowAmountRatio_y']) fbs['FlowAmountRatio'] = fbs['FlowAmountRatio'].fillna(0) # calculate flow amounts for each sector log.info("Calculating new flow amounts using flow ratios") fbs['FlowAmount'] = fbs['FlowAmount'] * fbs['FlowAmountRatio'] # drop columns log.info("Cleaning up new flow by sector") fbs = fbs.drop(columns=['Sector_x', 'FlowAmountRatio_x', 'Sector_y', 'FlowAmountRatio_y', 'FlowAmountRatio', 'ActivityProducedBy', 'ActivityConsumedBy']) # rename flow name to flowable fbs = fbs.rename(columns={"FlowName": 'Flowable', "Compartment": "Context" }) # drop rows where flowamount = 0 (although this includes dropping suppressed data) fbs = fbs[fbs['FlowAmount'] != 0].reset_index(drop=True) # add missing data columns fbs = add_missing_flow_by_fields(fbs, flow_by_sector_fields) # fill null values fbs = fbs.fillna(value=fbs_fill_na_dict) # aggregate df geographically, if necessary log.info("Aggregating flowbysector to " + method['target_geoscale'] + " level") if fips_number_key[v['geoscale_to_use']] < fips_number_key[attr['allocation_from_scale']]: from_scale = v['geoscale_to_use'] else: from_scale = attr['allocation_from_scale'] to_scale = method['target_geoscale'] fbs = agg_by_geoscale(fbs, from_scale, to_scale, fbs_default_grouping_fields, names) # aggregate data to every sector level log.info("Aggregating flowbysector to " + method['target_sector_level']) fbs = sector_aggregation(fbs, fbs_default_grouping_fields) # test agg by sector sector_agg_comparison = sector_flow_comparision(fbs) # return sector level specified in method yaml # load the crosswalk linking sector lengths cw = load_sector_length_crosswalk() sector_list = cw[method['target_sector_level']].unique().tolist() # add any non-NAICS sectors used with NAICS household = load_household_sector_codes() household = household.loc[household['NAICS_Level_to_Use_For'] == method['target_sector_level']] # add household sector to sector list sector_list.extend(household['Code'].tolist()) # subset df fbs = fbs.loc[(fbs[fbs_activity_fields[0]].isin(sector_list)) | (fbs[fbs_activity_fields[1]].isin(sector_list))].reset_index(drop=True) # add any missing columns of data and cast to appropriate data type fbs = add_missing_flow_by_fields(fbs, flow_by_sector_fields) log.info("Completed flowbysector for activity subset with flows " + ', '.join(map(str, names))) fbss.append(fbs) # create single df of all activities fbss = pd.concat(fbss, ignore_index=True, sort=False) # aggregate df as activities might have data for the same specified sector length fbss = aggregator(fbss, fbs_default_grouping_fields) # sort df fbss = fbss.sort_values( ['SectorProducedBy', 'SectorConsumedBy', 'Flowable', 'Context']).reset_index(drop=True) # save parquet file store_flowbysector(fbss, method_name)