Merge pull request #1102 from cal-itp/quarterly-metrics-q1

Quarterly performance metrics Q1 2024
cal-itp · May 8, 2024 · 4290b2c · 4290b2c
2 parents 85cfeaa + a3283ee
commit 4290b2c
Showing 12 changed files with 84 additions and 736 deletions.
diff --git a/_shared_utils/shared_utils/rt_dates.py b/_shared_utils/shared_utils/rt_dates.py
@@ -83,6 +83,8 @@ def get_week(month: Literal["apr2023", "oct2023"], exclude_wed: bool) -> list:
     "Q2_2023": "2023-04-12",
     "Q3_2023": "2023-07-12",
     "Q4_2023": "2023-10-11",
+    "Q1_2024": "2024-01-17",
+    "Q2_2024": "2024-04-17",
 }
 
 MONTH_DICT = {

diff --git a/portfolio/quarterly_performance_metrics/0__current_quarter_report__.ipynb b/portfolio/quarterly_performance_metrics/0__current_quarter_report__.ipynb
diff --git a/portfolio/quarterly_performance_metrics/0__historical_service_hours_v2__.ipynb b/portfolio/quarterly_performance_metrics/0__historical_service_hours_v2__.ipynb
diff --git a/quarterly_performance_objective/A1_scheduled_route_level_df.py b/quarterly_performance_objective/A1_scheduled_route_level_df.py
@@ -29,49 +29,20 @@ def shape_geom_to_route_geom(
     analysis_date: str
 ) -> gpd.GeoDataFrame: 
     """
-    Merge routes and trips to get line geometry.
+    Get longest shape for route-direction and tag
+    what category it is for parallel routes.
     """
-    cols_to_keep = ["route_id", "shape_id", "geometry"]
-    operator_cols = ["feed_key", "name"]
-
-    df = gtfs_schedule_wrangling.get_trips_with_geom(
-        analysis_date,
-        trip_cols = operator_cols + [ 
-                     "trip_id",
-                     "shape_id","shape_array_key", 
-                     "route_id", "route_type"
-                    ],
-        exclude_me = ["Amtrak Schedule", "*Flex"],
-        crs = geography_utils.CA_StatePlane
-        # make sure units are in feet
-    )
-
-    # Merge trips with the shape's line geom, and get it to shape_id level
-    df = (df[operator_cols + cols_to_keep + ["route_type"]]
-        .drop_duplicates(subset=operator_cols + ["shape_id"])
-        .reset_index(drop=True)
-    )
-
-    # Now, get it from shape_id level to route_id level
-    routes = create_parallel_corridors.process_transit_routes(
-        df, operator_cols = operator_cols 
-    ) 
-
-    # Add gtfs_dataset_name back
-    feed_gtfs_data_crosswalk = helpers.import_scheduled_trips(
-        analysis_date,
-        columns = ["feed_key", "gtfs_dataset_key"],
-        get_pandas = True
+    longest_shape = gtfs_schedule_wrangling.longest_shape_by_route_direction(
+        analysis_date
     ).rename(columns = {"schedule_gtfs_dataset_key": "gtfs_dataset_key"})
 
-    routes2 = pd.merge(
-        routes,
-        feed_gtfs_data_crosswalk,
-        on = "feed_key",
-        how = "inner"
+    operator_cols = ["feed_key", "gtfs_dataset_key"]
+
+    routes = create_parallel_corridors.process_transit_routes(
+        longest_shape, operator_cols = operator_cols 
     )
 
-    return routes2
+    return routes
 
 
 def aggregate_trip_service_to_route_level(
@@ -148,23 +119,24 @@ def add_district(route_df: gpd.GeoDataFrame,
 
     start = datetime.datetime.now()
 
-    route_cols = ["gtfs_dataset_key", "feed_key", "name", "route_id"]
+    route_cols_no_name = ["gtfs_dataset_key", "feed_key", "route_id"]
+    route_cols = route_cols_no_name + ["name"]
 
-    # Merge to get shape_level geometry, then pare down to route-level geometry
-    route_geom = shape_geom_to_route_geom(
+    # Get longest shape by route-direction
+    routes = shape_geom_to_route_geom(
         ANALYSIS_DATE)
-
-    # Add district
-    route_geom_with_district = add_district(route_geom, route_cols)
-
+    
+    # Add district    
+    route_geom_with_district = add_district(routes, route_cols_no_name)
+    
     # Get route-level service
     route_service = aggregate_trip_service_to_route_level(ANALYSIS_DATE, route_cols)
 
     # Merge service hours with route-level geometry and district
     gdf = pd.merge(
         route_geom_with_district,
         route_service,
-        on = route_cols,
+        on = route_cols_no_name,
         how = "left",
         validate = "1:1"
     )

diff --git a/quarterly_performance_objective/A4_add_route_speeds.py b/quarterly_performance_objective/A4_add_route_speeds.py
@@ -10,28 +10,24 @@
 from loguru import logger
 
 from calitp_data_analysis import utils
-from shared_utils import rt_utils, schedule_rt_utils
+from segment_speed_utils import gtfs_schedule_wrangling
+from segment_speed_utils.project_vars import SEGMENT_GCS, GTFS_DATA_DICT
 from update_vars import BUS_SERVICE_GCS, ANALYSIS_DATE
-from segment_speed_utils.project_vars import SEGMENT_GCS
-
 
 def aggregate_trip_speeds_to_route(
     analysis_date: str,
     route_cols: list
-):
+) -> pd.DataFrame:
     """
     Start with trip speeds and aggregate to route-level.
-    Instead of using route_speeds (which aggregates to 
-    route-direction-time_of_day and uses source_record_id),
-    let's just use trip-level speeds.
-    
-    Also, back out the operator's median speed, and use
-    that as the target speed for a given route.
     """
+    TRIP_SPEED_FILE = GTFS_DATA_DICT.rt_stop_times.trip_speeds_single_summary
+    MAX_SPEED = GTFS_DATA_DICT.rt_stop_times.max_speed
 
     speeds = pd.read_parquet(
-        f"{SEGMENT_GCS}trip_summary/trip_speeds_{analysis_date}.parquet",
-        columns = route_cols + ["trip_instance_key", "speed_mph"]
+        f"{SEGMENT_GCS}{TRIP_SPEED_FILE}_{analysis_date}.parquet",
+        columns = route_cols + ["trip_instance_key", "speed_mph"],
+        filters = [[("speed_mph", "<=", MAX_SPEED)]]
     )
 
     speeds_by_route = (speeds.groupby(route_cols, 
@@ -42,21 +38,6 @@ def aggregate_trip_speeds_to_route(
                        }).reset_index()
                        .rename(columns = {"trip_instance_key": "n_trips"})
                       )
-    '''
-    system_median = (speeds.groupby("schedule_gtfs_dataset_key", 
-                                    observed=True, group_keys=False)
-                     .agg({"speed_mph": "median"})
-                     .reset_index()
-                     .rename(columns = {"speed_mph": "system_speed_median"})
-                    )
-    
-    speeds_by_route2 = pd.merge(
-        speeds_by_route,
-        system_median,
-        on = "schedule_gtfs_dataset_key",
-        how = "inner"
-    )
-    '''
 
     return speeds_by_route
 
@@ -126,14 +107,9 @@ def speed_to_delay_estimate(
         indicator = True
     )
 
-    MERGE_CATEGORIES = {
-        "both": "rt_and_sched",
-        "left_only": "schedule_only",
-        "right_only": "rt_only"
-    }
-
     final = final.assign(
-        rt_sched_category = final._merge.map(MERGE_CATEGORIES)
+        rt_sched_category = final._merge.map(
+            gtfs_schedule_wrangling.sched_rt_category_dict)
     ).drop(columns = "_merge")
 
     TARGET_SPEEDS = {

diff --git a/quarterly_performance_objective/Makefile b/quarterly_performance_objective/Makefile
@@ -4,4 +4,4 @@ quarterly_performance_report:
 	python A2_generate_routes_on_shn_data.py
 	python A3_categorize_routes.py
 	python A4_add_route_speeds.py
-    
+	#cd ../ && make build_quarterly_performance_metrics -f Makefile