feat : add new analysis output (eqasim-org#266)

* feat : first analysis output tables * feat: add proportion & improvement outputs * fix : vehicle analysis & output folder * fix : coments & changelog * fix: correction with egt * fix: separate analysis from data output & update docs --------- Co-authored-by: Marie Laurent <[email protected]>
tellae · Nov 18, 2024 · d7453f4 · d7453f4
1 parent 2d3d9c1
commit d7453f4
Show file tree

Hide file tree

Showing 5 changed files with 151 additions and 5 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -2,6 +2,7 @@
 
 **Under development**
 
+- feat: add population analysis output
 - fix: avoid regenerating OSM when population changes
 - feat: add municipality information to households and activities
 - chore: update to `eqasim-java` commit `ece4932`

diff --git a/analysis/grid/comparison_flow_volume.py b/analysis/grid/comparison_flow_volume.py
@@ -1,8 +1,9 @@
 import pandas as pd
 import geopandas as gpd
+import os
 
 import plotly.express as px 
-
+ANALYSIS_FOLDER = "compare_flow_volume"
 
 SAMPLING_RATE = 0.05
 
@@ -84,6 +85,12 @@ def execute(context):
     df_grids = stat_grid(df_trips_comp,df_locations_comp,df_persons_comp,df_grid)
     point = df_grid.unary_union.centroid # a changé avec ploy_dep
     print("Printing grids...")
+
+    # check output folder existence
+    analysis_output_path = os.path.join(context.config("output_path"), ANALYSIS_FOLDER)
+    if not os.path.exists(analysis_output_path):
+        os.mkdir(analysis_output_path)    
+
     for prefix, figure in figures.items():
         df_select_age = df_stats[df_stats["age"].between(figure["min_age"],figure["max_age"])]
         df_select_age = df_select_age.dissolve(by=["id_carr_1km","following_purpose"],aggfunc="count").reset_index()
@@ -103,14 +110,14 @@ def execute(context):
                 df_select  = df_select[df_select["count"] != 0]
                 fig = px.choropleth_mapbox(df_select,geojson=df_select.geometry,locations=df_select.index,color="count", opacity= 0.7,color_continuous_scale='reds',
                                         mapbox_style = 'open-street-map',center=dict(lat= point.y,lon=point.x),title=f"Localisation flow distribution for {prefix} group with {purpose} purpose")
-                fig.write_html(f'{context.config("output_path")}/{context.config("output_prefix")}{prefix}_{purpose}.html')
+                fig.write_html(f'{analysis_output_path}/{context.config("output_prefix")}{prefix}_{purpose}.html')
             else :
                 df_grids_select = gpd.sjoin(df_grids_select,df_grid,how='right',predicate="contains").fillna(0)
                 df_select = gpd.sjoin(df_select,df_grids_select.drop(columns=[ 'index_left']),how='right',predicate="contains").rename(columns={"count_left":"volume_studied_simu","count_right":"volume_compared_simu"}).fillna(0)
                 df_select["volume_difference"] = df_select["volume_studied_simu"] - df_select["volume_compared_simu"]
                 df_select  = df_select[(df_select["volume_studied_simu"] != 0 )| (df_select["volume_compared_simu"] != 0)]
                 df_select["pourcentage_vol"] = df_select["volume_difference"] / df_select["volume_compared_simu"]
                 px.choropleth_mapbox(df_select,geojson=df_select.geometry,locations=df_select.index,color="volume_difference", opacity= 0.7,color_continuous_scale="picnic", color_continuous_midpoint= 0,hover_name="id_carr_1km_right", hover_data=["volume_studied_simu", "volume_compared_simu","pourcentage_vol"],
-                                        mapbox_style = 'open-street-map',center=dict(lat= point.y,lon=point.x),title=f"Comparison flow distribution with previous simulation for {prefix} group with {purpose} purpose").write_html(f'{context.config("output_path")}/{context.config("output_prefix")}{prefix}_{purpose}.html')
+                                        mapbox_style = 'open-street-map',center=dict(lat= point.y,lon=point.x),title=f"Comparison flow distribution with previous simulation for {prefix} group with {purpose} purpose").write_html(f'{analysis_output_path}/{context.config("output_prefix")}{prefix}_{purpose}.html')
 
 
diff --git a/analysis/synthesis/population.py b/analysis/synthesis/population.py
@@ -0,0 +1,127 @@
+
+import os
+import numpy as np
+import pandas as pd
+import geopandas as gpd
+from  analysis.marginals import NUMBER_OF_VEHICLES_LABELS
+from shapely import distance
+AGE_CLASS = [0, 10, 14, 17, 25, 50, 65, np.inf]
+NUMBER_OF_VEHICLES= [0,1,2,3,np.inf]
+NAME_AGE_CLASS = ["0-10","11-14","15-17","18-25","26-50","51-65","65+"]
+ANALYSIS_FOLDER = "analysis_population"
+def configure(context):
+
+    context.config("output_path")
+    context.config("output_prefix", "ile_de_france_")
+    context.config("sampling_rate")
+
+    context.stage("synthesis.population.trips")
+    context.stage("synthesis.population.enriched")
+    context.stage("synthesis.population.spatial.locations")
+
+    context.stage("data.census.filtered", alias = "census")
+    context.stage("data.hts.selected", alias = "hts")
+
+def execute(context):
+
+    # check output folder existence
+    analysis_output_path = os.path.join(context.config("output_path"), ANALYSIS_FOLDER)
+    if not os.path.exists(analysis_output_path):
+        os.mkdir(analysis_output_path)
+
+    prefix = context.config("output_prefix")
+    sampling_rate = context.config("sampling_rate")
+    df_person_eq = context.stage("synthesis.population.enriched")
+    df_trip_eq = context.stage("synthesis.population.trips")
+    df_location_eq = context.stage("synthesis.population.spatial.locations")[["person_id", "activity_index", "geometry"]]
+
+    df_location_eq = df_location_eq.to_crs("EPSG:2154")
+    df_trip_eq["preceding_activity_index"] = df_trip_eq["trip_index"]
+    df_trip_eq["following_activity_index"] = df_trip_eq["trip_index"] + 1
+
+    df_census = context.stage("census")
+    df_hts_households, df_hts_person, df_hts_trip = context.stage("hts")
+    df_hts_person["person_weight"] *=df_census["weight"].sum()/df_hts_person["person_weight"].sum()
+    df_hts_households["household_weight"] *=df_census["weight"].sum()/df_hts_households["household_weight"].sum()
+    # get age class
+    df_person_eq["age_class"] = pd.cut(df_person_eq["age"],AGE_CLASS,include_lowest=True,labels=NAME_AGE_CLASS)
+    df_census["age_class"] = pd.cut(df_census["age"],AGE_CLASS,include_lowest=True,labels=NAME_AGE_CLASS)
+    df_hts_person["age_class"] = pd.cut(df_hts_person["age"],AGE_CLASS,include_lowest=True,labels=NAME_AGE_CLASS)
+
+    # get vehicule class 
+    df_person_eq["vehicles_class"] = pd.cut(df_person_eq["number_of_vehicles"],NUMBER_OF_VEHICLES,right=True,labels=NUMBER_OF_VEHICLES_LABELS)
+    df_hts_households["vehicles_class"] = pd.cut(df_hts_households["number_of_vehicles"],NUMBER_OF_VEHICLES,right=True,labels=NUMBER_OF_VEHICLES_LABELS)
+
+
+    df_eq_travel = pd.merge(df_trip_eq,df_person_eq[["person_id","age_class"]],on=["person_id"])
+    df_hts_travel = pd.merge(df_hts_trip,df_hts_person[["person_id","age_class","person_weight"]],on=["person_id"])
+
+    print("Generate tables ...")
+    # Age purpose analysis
+    analysis_age_purpose = pd.pivot_table(df_eq_travel,"person_id",index="age_class",columns="following_purpose",aggfunc="count")
+    analysis_age_purpose = analysis_age_purpose/sampling_rate
+    analysis_age_purpose.to_csv(f"{analysis_output_path}/{prefix}age_purpose.csv")
+
+    # Compare age volume
+    analysis_age_class = pd.concat([df_census.groupby("age_class")["weight"].sum(),df_person_eq.groupby("age_class")["person_id"].count()],axis=1).reset_index()
+    analysis_age_class.columns = ["Age class","INSEE","EQASIM"]
+    analysis_age_class["Proportion_INSEE"] = analysis_age_class["INSEE"] /df_census["weight"].sum()
+    analysis_age_class["Proportion_EQASIM"] = analysis_age_class["EQASIM"] /len(df_person_eq)
+    analysis_age_class["EQASIM"] = analysis_age_class["EQASIM"]/sampling_rate
+    analysis_age_class.to_csv(f"{analysis_output_path}/{prefix}age.csv")
+
+    # Compare vehicle volume
+    analysis_vehicles_class = pd.concat([df_hts_households.groupby("vehicles_class")["household_weight"].sum(),df_person_eq.groupby("vehicles_class")["household_id"].nunique()],axis=1).reset_index()
+    analysis_vehicles_class.columns = ["Number of vehicles class","HTS","EQASIM"]
+    analysis_vehicles_class["Proportion_HTS"] = analysis_vehicles_class["HTS"] / df_hts_households["household_weight"].sum() 
+    analysis_vehicles_class["Proportion_EQASIM"] = analysis_vehicles_class["EQASIM"] / df_person_eq["household_id"].nunique() 
+    analysis_vehicles_class.to_csv(f"{analysis_output_path}/{prefix}nbr_vehicle.csv")
+
+    # Compare license volume 
+    analysis_license_class = pd.concat([df_hts_person.groupby("has_license")["person_weight"].sum(),df_person_eq.groupby("has_license")["person_id"].count()],axis=1).reset_index()
+    analysis_license_class.columns = ["Possession of license","HTS","EQASIM"]
+    analysis_license_class["Proportion_HTS"] = analysis_license_class["HTS"] /df_hts_person["person_weight"].sum()
+    analysis_license_class["Proportion_EQASIM"] = analysis_license_class["EQASIM"] /len(df_person_eq)
+    analysis_license_class["EQASIM"] = analysis_license_class["EQASIM"]/sampling_rate
+    analysis_license_class.to_csv(f"{analysis_output_path}/{prefix}license.csv")
+
+    # Compare travel volume
+    analysis_travel = pd.concat([df_hts_travel.groupby("age_class")["person_weight"].sum(),df_eq_travel.groupby("age_class")["person_id"].count()],axis=1).reset_index()
+    analysis_travel.columns = ["Age class","HTS","EQASIM"]
+    analysis_travel["Proportion_HTS"] = analysis_travel["HTS"] /df_hts_travel["person_weight"].sum()
+    analysis_travel["Proportion_EQASIM"] = analysis_travel["EQASIM"] /len(df_eq_travel)
+    analysis_travel["EQASIM"] = analysis_travel["EQASIM"]/sampling_rate
+    analysis_travel.to_csv(f"{analysis_output_path}/{prefix}travel.csv")
+
+    # Compare distance
+    df_hts_travel["routed_distance"] = df_hts_travel["routed_distance"]/1000 if "routed_distance" in  df_hts_travel.columns else df_hts_travel["euclidean_distance"]/1000
+    df_hts_travel["distance_class"] = pd.cut(df_hts_travel["routed_distance"],list(np.arange(100))+[np.inf])
+
+    df_spatial = pd.merge(df_trip_eq, df_location_eq.rename(columns = {
+        "activity_index": "preceding_activity_index",
+        "geometry": "preceding_geometry"
+    }), how = "left", on = ["person_id", "preceding_activity_index"])
+
+    df_spatial = pd.merge(df_spatial, df_location_eq.rename(columns = {
+        "activity_index": "following_activity_index",
+        "geometry": "following_geometry"
+    }), how = "left", on = ["person_id", "following_activity_index"])
+    df_spatial["distance"] = df_spatial.apply(lambda x:distance( x["preceding_geometry"],x["following_geometry"])/1000,axis=1)
+    df_spatial["distance_class"] = pd.cut(df_spatial["distance"],list(np.arange(100))+[np.inf])
+
+    analysis_distance = pd.concat([df_hts_travel.groupby("distance_class")["person_weight"].sum(),df_spatial.groupby("distance_class")["person_id"].count()],axis=1).reset_index()
+    analysis_distance.columns = ["Distance class","HTS","EQASIM"]
+    analysis_distance["Proportion_HTS"] = analysis_distance["HTS"] / analysis_distance["HTS"].sum()
+    analysis_distance["Proportion_EQASIM"] = analysis_distance["EQASIM"] / len(df_spatial)
+    analysis_distance["EQASIM"] = analysis_distance["EQASIM"]/ sampling_rate
+    analysis_distance.to_csv(f"{analysis_output_path}/{prefix}distance.csv")
+
+
+
+
+
+
+
+
+
+
diff --git a/docs/population.md b/docs/population.md
@@ -450,3 +450,11 @@ folder as: `{output_prefix}_{age group}_{trip pupose}.html`
 
 Note:
 With `analysis_from_file` at False, the last synthetic population is studied by default. Also if `output_prefix` and `comparison_file_prefix` refer to the same outputs, or `comparison_file_prefix` is not specified, then only a volume visualisation of this particular population is produced.
+
+
+### Comparaison population to source data
+
+Using the population pipeline in the Analysis directory, you can generate multiple tables comparing composition of synthetic population to source data. Right now the tables generated compare : population volume by age range, households volume by number of vehicles, population volume with a license and without, trip volume by age range and trip volume by length.
+Complementary from the synthetic population only, a table of population volume by age range and trip purpose is also created.
+
+To be able to use this pipeline, you must already have create a synthetic population. Then you need to open the `config.yml` and add the `analysis.synthesis.population` stage in the `run` section. 
diff --git a/synthesis/output.py b/synthesis/output.py
@@ -6,8 +6,10 @@
 import sqlite3
 import math
 import numpy as np
+from analysis.synthesis.population import ANALYSIS_FOLDER
 
 def configure(context):
+
     context.stage("synthesis.population.enriched")
 
     context.stage("synthesis.population.activities")
@@ -22,7 +24,8 @@ def configure(context):
     context.config("output_path")
     context.config("output_prefix", "ile_de_france_")
     context.config("output_formats", ["csv", "gpkg"])
-
+    context.config("sampling_rate")
+
     if context.config("mode_choice", False):
         context.stage("matsim.simulation.prepare")
 
@@ -270,4 +273,4 @@ def execute(context):
         clean_gpkg(path)
     if "geoparquet" in output_formats:
         path = "%s/%strips.geoparquet" % (output_path, output_prefix)
-        df_spatial.to_parquet(path)
+        df_spatial.to_parquet(path)