Merge pull request #1209 from cal-itp/speed-aggregation-ts

segment line-up as time-series

Author: tiffanychu90

rt_segment_speeds/39_segment_stability.ipynb

@@ -0,0 +1,57 @@
+import geopandas as gpd
+import pandas as pd
+
+from segment_speed_utils.project_vars import RT_SCHED_GCS, GTFS_DATA_DICT
+
+route_dir_stop_cols = [
+    "schedule_gtfs_dataset_key", "name", "time_period",
+    "route_id", "direction_id", 
+    "stop_pair", "stop_pair_name"
+]
+
+operator_list = df = pd.read_parquet(
+    f"{RT_SCHED_GCS}{GTFS_DATA_DICT.digest_tables.route_segment_speeds}.parquet",
+    columns = ["name"]
+).name.unique()
+
+def route_segment_speeds_ts(
+    file: str = GTFS_DATA_DICT.digest_tables.route_segment_speeds,
+    **kwargs
+) -> gpd.GeoDataFrame:
+    """
+    Subset the concatenated time-series route-segment speeds
+    and grab all-day only and a smaller set of columns.
+    """
+    df = gpd.read_parquet(
+        f"{RT_SCHED_GCS}{file}.parquet",
+        **kwargs
+        #filters = [[("time_period", "==", "all_day")]],
+        #columns = route_dir_stop_cols + ["service_date", "p50_mph", "geometry"]
+    )
+    
+    return df
+
+
+def count_time_series_values_by_route_direction_stop(
+    df: gpd.GeoDataFrame,
+    group_cols: list
+) -> pd.DataFrame:
+    """
+    For each stop, count how many segment variations are available
+    across our time-series.
+    """
+    df2 = (df
+           .groupby(group_cols, group_keys=False)
+           .agg({
+                "p50_mph": "count",
+                "service_date": "nunique",
+                "geometry": "nunique"
+            }).reset_index()
+           .rename(columns = {
+                "p50_mph": "n_speed_values",
+                "service_date": "n_dates",
+                "geometry": "n_geometry"
+            })
+          )
+    
+    return df2