import geopandas as gpd import pandas as pd import json import os import matplotlib.pyplot as plt # Folder containing all simulated plans: # map_0.json, map_1.json, ..., map_5000.json MAPS_FOLDER = "/Users/edieneuville/Desktop/graph decomp/SEPFILE/Simulation" # Output files OUTPUTCSV = "/Users/edieneuville/Desktop/graph decomp/SEPFILE/ensemble_results.csv" SEATDISTCSV = "/Users/edieneuville/Desktop/graph decomp/SEPFILE/seat_distributions.csv" MMDISTCSV = "/Users/edieneuville/Desktop/graph decomp/SEPFILE/majority_minority_districts.csv" MMTCSV = "/Users/edieneuville/Desktop/graph decomp/SEPFILE/majority_minority_counts_per_map.csv" #elections to analyze ELECTIONS = [ ("PRES12D", "PRES12R", "PRES12"), ("GOV12D", "GOV12R", "GOV12"), ("SEN12D", "SEN12R", "SEN12"), ("USH12D", "USH12R", "USH12"), ("GOV14D", "GOV14R", "GOV14"), ("USH14D", "USH14R", "USH14"), ("PRES16D", "PRES16R", "PRES16"), ("SEN16D", "SEN16R", "SEN16"), ("USH16D", "USH16R", "USH16"), ("GOV18D", "GOV18R", "GOV18"), ("SEN18D", "SEN18R", "SEN18"), ("USH18D", "USH18R", "USH18"), ] # The JSON maps only contain the district assignment for each ward wards = gpd.read_file("WI.shp") # Reset index so row numbers match the node IDs used in JSON maps wards = wards.reset_index(drop=True) print("Loaded shapefile") print("Rows:", len(wards)) #Create a node_id column that matches the JSON keys wards["node_id"] = wards.index.astype(str) #Majority Minority districts counts #(TOTPOP - NH_WHITE) > NH_WHITE wards["NONWHITE_POP"] = wards["TOTPOP"] - wards["NH_WHITE"] wards["MAJORITY_MINORITY_WARD"] = wards["NONWHITE_POP"] > wards["NH_WHITE"] #count statewide majority-minority wards num_mm_wards = int(wards["MAJORITY_MINORITY_WARD"].sum()) def count_district_wins(gdf, dem_col, rep_col, district_col="district"): #Aggregate ward-level votes up to districts, then count how many #districts Democrats and Republicans win #dem_wins = number of districts Democrats win #rep_wins = number of districts Republicans win #ties = number of tied districts #grouped = district-level vote totals #Sum vote totals within each district grouped = gdf.groupby(district_col)[[dem_col, rep_col]].sum() #Count district wins dem_wins = int((grouped[dem_col] > grouped[rep_col]).sum()) rep_wins = int((grouped[rep_col] > grouped[dem_col]).sum()) ties = int((grouped[dem_col] == grouped[rep_col]).sum()) return dem_wins, rep_wins, ties, grouped map_files = sorted( [f for f in os.listdir(MAPS_FOLDER) if f.endswith(".json")] ) #These lists will collect all output rows results = [] majority_minority_rows = [] #load the JSON #assign district labels to the shapefile rows #compute election outcomes by district #compute majority-minority districts #save one summary row for that map for idx, filename in enumerate(map_files): if idx % 100 == 0: print(f"Processing map {idx} / {len(map_files)}") path = os.path.join(MAPS_FOLDER, filename) # Load one simulated map with open(path, "r") as f: map_data = json.load(f) #copy the shapefile data so we can assign this map's districts gdf = wards.copy() # Replace district labels using the JSON gdf["district"] = gdf["node_id"].map(lambda x: map_data[x]["district"]) #start one summary row for this map row_result = { "map_file": filename, "majority_minority_wards_statewide": num_mm_wards } for dem_col, rep_col, label in ELECTIONS: total_dem_votes = int(wards[dem_col].sum()) total_rep_votes = int(wards[rep_col].sum()) row_result[f"{label}_D_total_votes"] = total_dem_votes row_result[f"{label}_R_total_votes"] = total_rep_votes # Aggregate TOTPOP and NH_WHITE to district level district_demo = ( gdf.groupby("district")[["TOTPOP", "NH_WHITE"]] .sum() .reset_index() ) district_demo["NONWHITE_POP"] = district_demo["TOTPOP"] - district_demo["NH_WHITE"] district_demo["MAJORITY_MINORITY_DISTRICT"] = district_demo["NONWHITE_POP"] > district_demo["NH_WHITE"] # Save each majority-minority district for this map for _, row in district_demo.iterrows(): if row["MAJORITY_MINORITY_DISTRICT"]: majority_minority_rows.append({ "map_file": filename, "district": int(row["district"]), "TOTPOP": int(row["TOTPOP"]), "NH_WHITE": int(row["NH_WHITE"]), "NONWHITE_POP": int(row["NONWHITE_POP"]) }) # Also save how many majority-minority districts this map has row_result["num_majority_minority_districts"] = int( district_demo["MAJORITY_MINORITY_DISTRICT"].sum() ) # For each election, count Democratic and Republican district wins for dem_col, rep_col, label in ELECTIONS: dem_wins, rep_wins, ties, grouped = count_district_wins(gdf, dem_col, rep_col) # Save seat counts into the results row row_result[f"{label}_D_wins"] = dem_wins row_result[f"{label}_R_wins"] = rep_wins row_result[f"{label}_ties"] = ties # Optional: average Democratic vote share across districts vote_share = grouped[dem_col] / (grouped[dem_col] + grouped[rep_col]) row_result[f"{label}_D_vote_share_mean"] = vote_share.mean() # Store this map's full summary row results.append(row_result) #Save results to CSV results_df = pd.DataFrame(results) results_df.to_csv(OUTPUTCSV, index=False) #Saving majority-minority district details and counts #which map had a majority-minority district #which district it was #its district-level population totals majority_minority_df = pd.DataFrame(majority_minority_rows) majority_minority_df.to_csv(MMDISTCSV , index=False) #also save a simpler file: how many majority-minority districts each map had if not majority_minority_df.empty: mm_count_per_map = ( majority_minority_df.groupby("map_file") .size() .reset_index(name="num_majority_minority_districts") ) else: # If none exist, still write an empty summary using the map list mm_count_per_map = pd.DataFrame({ "map_file": map_files, "num_majority_minority_districts": [0] * len(map_files) }) mm_count_per_map.to_csv(MMTCSV, index=False) print("Saved majority-minority district counts per map to:") print(MMTCSV ) # For each election: #count how many maps gave Democrats k seats #count how many maps gave Republicans k seats #save those distributions #plot histograms seat_distribution_rows = [] # Number of districts is the max district label count in any map. # Since you are working with congressional plans, this should be 8. numDistricts = int(results_df[[f"{label}_D_wins" for _, _, label in ELECTIONS]].max().max() + results_df[[f"{label}_R_wins" for _, _, label in ELECTIONS]].max().max()) # The previous line overcounts because D wins + R wins = seats, so instead # just use the maximum D wins or R wins seen, and compare against known 8 if needed. numDistricts = max( int(results_df[f"{label}_D_wins"].max()) for _, _, label in ELECTIONS ) numDistricts = max( numDistricts, max(int(results_df[f"{label}_R_wins"].max()) for _, _, label in ELECTIONS) ) # If you know you always have 8 districts, you can just do: # numDistricts = 8 for _, _, label in ELECTIONS: print(f"{label} seat distribution") # Count how many maps gave Democrats k seats dem_counts = ( results_df[f"{label}_D_wins"] .value_counts() .sort_index() ) # Count how many maps gave Republicans k seats rep_counts = ( results_df[f"{label}_R_wins"] .value_counts() .sort_index() ) # Convert to full arrays from 0 to numDistricts dem_array = [dem_counts.get(k, 0) for k in range(numDistricts + 1)] rep_array = [rep_counts.get(k, 0) for k in range(numDistricts + 1)] print("Democratic seat count array:") print(dem_array) print("Republican seat count array:") print(rep_array) # Save one long-format row for every possible seat count for seats in range(numDistricts + 1): seat_distribution_rows.append({ "election": label, "party": "D", "seats_won": seats, "num_maps": dem_array[seats] }) seat_distribution_rows.append({ "election": label, "party": "R", "seats_won": seats, "num_maps": rep_array[seats] }) # Democratic seat histogram plt.figure(figsize=(8, 5)) plt.hist( results_df[f"{label}_D_wins"], bins=range(0, numDistricts + 2), align="left", rwidth=0.8 ) plt.xticks(range(0, numDistricts + 1)) plt.xlabel("Democratic districts won") plt.ylabel("Number of maps") plt.title(f"{label} Democratic Seat Distribution") plt.show() # Republican seat histogram plt.figure(figsize=(8, 5)) plt.hist( results_df[f"{label}_R_wins"], bins=range(0, numDistricts + 2), align="left", rwidth=0.8 ) plt.xticks(range(0, numDistricts + 1)) plt.xlabel("Republican districts won") plt.ylabel("Number of maps") plt.title(f"{label} Republican Seat Distribution") plt.show() #seat dists seat_dist_df = pd.DataFrame(seat_distribution_rows) seat_dist_df.to_csv(SEATDISTCSV , index=False) print("\nSaved seat distributions to:") print(SEATDISTCSV )