World Cup Transit Service near SoFi Stadium

import warnings
warnings.filterwarnings("ignore")
                        
import altair as alt
import branca.colormap as cm
import folium
import geopandas as gpd
import gcsfs
import google.auth
import pandas as pd

from dotenv import dotenv_values
from IPython.display import HTML, display

import world_cup_vars as wc_vars
import event_helpers
import create_route_and_stop_event_summary as prep_data
import chart_utils
import poi_geography_utils as poi_geog_utils

credentials, _ = google.auth.default()
GCS_FILE_PATH = wc_vars.GCS_FILE_PATH
CARTO_TILE_URL = dotenv_values("../_env")["CARTO_TILE_URL"]

STADIUM_NAME = "sofi"
STADIUM_FULL_NAME = "SoFi Stadium"

alt.data_transformers.enable("vegafusion")

# Set drop down menu to be on the upper right for the charts
display(
    HTML(
        """
<style>
form.vega-bindings {
  position: absolute;
  right: 0px;
  top: 0px;
}
</style>
"""
    )
)

# daily schedule + RT route-direction summary
daily_trips = prep_data.prep_fct_daily_schedule_rt_route_direction_summary(
    wc_vars.event_name,
    wc_vars.socal_names,
    wc_vars.sofi_match_times,
)

# special routes are tagged, filter for these and merge 
routes_near_poi = prep_data.import_routes_near_poi(
    wc_vars.event_name, STADIUM_NAME
).query(
    'is_special_route==True'
)

sofi_trips = pd.merge(
    daily_trips,
    routes_near_poi,
    on = ["feed_key", "schedule_name", "route_id", "direction_id", "route_name"],
    how = "inner"
)

daily_trips_by_operator = event_helpers.aggregate_daily_trips(
    sofi_trips, ["service_date", "schedule_name"])

# just add the before and after, set to zero, otherwise LA Metro Events 
# single point drops away from chart visibility
append_la_metro_events = pd.DataFrame({
    'service_date': ["2026-07-09", "2026-07-11"],
})
append_la_metro_events = append_la_metro_events.assign(
    service_date = pd.to_datetime(append_la_metro_events.service_date).dt.normalize(),
    schedule_name = "LA Metro Events Schedule",
    n_trips = 0
)

daily_trips_by_operator2 = pd.concat(
    [daily_trips_by_operator, append_la_metro_events], 
    axis=0, ignore_index=True
)

chart_utils.trip_chart_with_event_dates(
    daily_trips_by_operator2, wc_vars.sofi_dates, color_col="schedule_name"
).properties(
    title= "Daily Trips by Operator during World Cup",
    width=500, height=300
)

daily_trips_by_route = event_helpers.aggregate_daily_trips(
    sofi_trips, ["service_date", "schedule_name", "route_name"]
)

chart_utils.trip_chart_with_event_dates(
    daily_trips_by_route, wc_vars.sofi_dates, color_col="route_name"
).properties(
    title= "Daily Trips by Route during World Cup",
    width=500, height=300
)

# Get each route's daily average number of trips on event days vs non-event days - plot on map
route_gdf = gpd.read_parquet(
    f"{GCS_FILE_PATH}route_summary_{STADIUM_NAME}.parquet",
    storage_options = {"token": credentials.token}
)

stop_gdf = gpd.read_parquet(
    f"{GCS_FILE_PATH}stop_summary_{STADIUM_NAME}.parquet",
    storage_options = {"token": credentials.token}
)

poi = gpd.read_parquet(
    f"{wc_vars.GCS_FILE_PATH}points_of_interest_{wc_vars.event_name}.parquet",
    storage_options = {"token": credentials.token},
    filters = [[("point_of_interest", "==", STADIUM_FULL_NAME)]]
)

# check that our change column works with the shared legend cutoff across the maps in this notebook
#route_gdf.combined_change_daily_trips.min(), route_gdf.combined_change_daily_trips.max()

# check that our change column works with the shared legend cutoff across the maps in this notebook
#stop_gdf.combined_change_daily_arrivals.min(), stop_gdf.combined_change_daily_arrivals.max()

from _color_palette import SERVICE_CHANGE_COLORS, SERVICE_CHANGE_CAPTION
SERVICE_CHANGE_INDEX = [-200, -100, -1, 1, 50, 100, 200, 300, 800]

route_map = route_gdf[
    (route_gdf.schedule_name!="LA Metro Events Schedule") & 
    ((route_gdf.is_route_near == True) | (route_gdf.is_special_route == True))
    ].explore(
    "combined_change_daily_trips",
    tiles = CARTO_TILE_URL,
    attr="(C) OpenStreetMap contributors (C) CARTO",
    cmap = cm.StepColormap(
        colors=SERVICE_CHANGE_COLORS, 
        index=SERVICE_CHANGE_INDEX, 
        vmin=route_gdf.combined_change_daily_trips.min(), 
        vmax=route_gdf.combined_change_daily_trips.max(), 
        tick_labels=SERVICE_CHANGE_INDEX,
        caption=SERVICE_CHANGE_CAPTION
    ),
    style_kwds = {
        "weight": 3
    },
    name = "Additional Trips (compared to baseline) on Special Service Routes" 
)

# plot LA Metro events feed separately
# sometimes the route geom will overlap exactly
route_map = route_gdf[(route_gdf.schedule_name=="LA Metro Events Schedule")].explore(
    "combined_change_daily_trips",
    m = route_map,
    cmap = cm.StepColormap(
        colors=SERVICE_CHANGE_COLORS, 
        index=SERVICE_CHANGE_INDEX, 
        vmin=route_gdf.combined_change_daily_trips.min(), 
        vmax=route_gdf.combined_change_daily_trips.max(), 
        tick_labels=SERVICE_CHANGE_INDEX,
        caption=SERVICE_CHANGE_CAPTION
    ),
    style_kwds = {
        "weight": 3
    },
    name = "LA Metro Special World Cup Service" 
)

# keep stops that are on routes that got near 
route_map = stop_gdf[
    (stop_gdf.is_route_near == True) | (stop_gdf.is_special_route == True)
].drop(columns = ["near_route_ids", "special_route_ids"]).explore(
    "combined_change_daily_arrivals",
    m = route_map,
    cmap = cm.StepColormap(
        colors=SERVICE_CHANGE_COLORS, 
        index=SERVICE_CHANGE_INDEX, 
        vmin=stop_gdf.combined_change_daily_arrivals.min(), 
        vmax=stop_gdf.combined_change_daily_arrivals.max(), 
        tick_labels=SERVICE_CHANGE_INDEX,
        caption=SERVICE_CHANGE_CAPTION
    ),
    style_kwds={
        "style_function": lambda x: {
            "radius": x["properties"]["combined_change_daily_arrivals"]*0.05,
            "fillOpacity": 0.5,
            #"color": "#808080", # stroke color, if defined, stroke color will not be the color of the marker
            "strokeOpacity": 1,
            "weight": 2,
            #"weight":x["properties"]["combined_change_daily_arrivals"]*0.08, # weight seems to be stroke weight
        }
    },
    name = "Additional Arrivals (compared to baseline)" 
)

# https://python-visualization.github.io/folium/latest/user_guide/geojson/geojson_marker.html
# available colors, https://python-visualization.github.io/folium/latest/reference.html#folium.map.Marker
route_map = poi.explore(
    "point_of_interest",
    m=route_map,
    marker_type = "marker",
    name=STADIUM_FULL_NAME,
    legend = False, # legend color is blue no matter what color is set for icon, confusing
    marker_kwds=dict(icon=folium.Icon(icon="star", color="darkpurple")),
)

folium.LayerControl().add_to(route_map)
route_map

la_metro_events = sofi_trips[sofi_trips.schedule_name == "LA Metro Events Schedule"].reset_index(drop=True)

#https://matplotlib.org/stable/users/explain/colors/colormaps.html
la_metro_events[["route_name", "direction_id", "geometry", "n_trips"]].explore(
    "route_name",
    tiles = CARTO_TILE_URL,
    attr="(C) OpenStreetMap contributors (C) CARTO",
    style_kwds={"style_function": lambda x: {"weight":x["properties"]["n_trips"]*0.05}},
    cmap="Set1"
)

# filter to stops near stadium
stops_near_gdf = stop_gdf[(stop_gdf.is_stop_near == True) & 
    (stop_gdf.is_special_route == True)]

operator_df = (
    stops_near_gdf
    .groupby(["schedule_name", "route_id_array", "stop_name"])
    .agg({
        "change_daily_arrivals_weekday": "sum",
        "change_daily_arrivals_weekend": "sum",
        "stop_id": "nunique"
    })
    .reset_index()
    .rename(columns = {"stop_id": "n_stop_ids"})
)

for i in sorted(operator_df.schedule_name.unique()):
    chart = chart_utils.weekday_weekend_chart_by_operator(operator_df, i)
    display(chart)

stop_tod_df = pd.read_parquet(
    f"{GCS_FILE_PATH}arrivals_time_of_day_summary_{STADIUM_NAME}.parquet",
    filesystem = gcsfs.GCSFileSystem()
)

stop_tod_near = poi_geog_utils.keep_stops_within_buffer(
    stop_gdf,
    stop_tod_df,
    poi,
)

stop_tod_near2 = (
    stop_tod_near
    .groupby(["service_date", "schedule_name", "time_of_day", "day_type", "route_id_array"], dropna=False)
    .agg({
        "arrivals_per_hour": "mean",
        "stop_id": "count"
    }).reset_index().rename(columns = {"stop_id": "n_stops"})
)

stop_tod_near2 = stop_tod_near2.assign(
    arrivals_per_hour = stop_tod_near2.arrivals_per_hour.round(2)
)

list_of_operators = sorted(stop_tod_near2.schedule_name.unique())

MATCH_TOD = "midday"
MATCH_DAY_TYPE = "weekday"

filtered_sofi_dict = event_helpers.grab_matches_by_day_type(
    wc_vars.sofi_match_times, MATCH_TOD, MATCH_DAY_TYPE)

MATCH_DAYS = list(filtered_sofi_dict.keys())

operator_dropdown = alt.binding_select(
    options=list_of_operators,
    name="GTFS Schedule: ",
)

dropdown_param = alt.selection_point(
    fields=["schedule_name"], value=list_of_operators[0], bind=operator_dropdown
)

selection = alt.selection_point(fields=["route_id_array"], bind="legend")

indiv_time_of_day_charts = [
    chart_utils.arrivals_by_time_of_day_grouped_bar(
        stop_tod_near2[stop_tod_near2.service_date.isin(["nonevent"] + MATCH_DAYS)], 
        time_of_day, 
        MATCH_DAY_TYPE
    ).properties(
        title = f"{time_of_day}", width=200, height=100
    ).add_params(selection, dropdown_param).transform_filter(dropdown_param)
    for time_of_day in ["am_peak", "midday", "pm_peak", "early_am", "evening", "owl"]
    # set time-of-day order to put times surrounding match at the top
]    

alt.vconcat(
    alt.hconcat(*indiv_time_of_day_charts[:3]).resolve_scale(y ='shared'),
    alt.hconcat(*indiv_time_of_day_charts[3:]).resolve_scale(y ='shared')
).add_params(selection, dropdown_param).properties(
    title={
        "text": f"Sofi {MATCH_DAY_TYPE.title()} Matches on {', '.join(MATCH_DAYS)}", 
        "subtitle": [f"Match Starts: {MATCH_TOD}", "Travel Changes: am_peak, midday, pm_peak"]
}).resolve_scale(y ='shared')

MATCH_TOD = "midday"
MATCH_DAY_TYPE = "weekend"

filtered_sofi_dict = event_helpers.grab_matches_by_day_type(
    wc_vars.sofi_match_times, MATCH_TOD, MATCH_DAY_TYPE)

MATCH_DAYS = list(filtered_sofi_dict.keys())

operator_dropdown = alt.binding_select(
    options=list_of_operators,
    name="GTFS Schedule: ",
)

dropdown_param = alt.selection_point(
    fields=["schedule_name"], value=list_of_operators[0], bind=operator_dropdown
)

selection = alt.selection_point(fields=["route_id_array"], bind="legend")

indiv_time_of_day_charts = [
    chart_utils.arrivals_by_time_of_day_grouped_bar(
        stop_tod_near2[stop_tod_near2.service_date.isin(["nonevent"] + MATCH_DAYS)], 
        time_of_day, 
        MATCH_DAY_TYPE
    ).properties(
        title = f"{time_of_day}", width=200, height=100
    ).add_params(selection, dropdown_param).transform_filter(dropdown_param) 
    for time_of_day in ["am_peak", "midday", "pm_peak", "early_am", "evening", "owl"]
    # set time-of-day order to put times surrounding match at the top
]    

alt.vconcat(
    alt.hconcat(*indiv_time_of_day_charts[:3]).resolve_scale(y ='shared'),
    alt.hconcat(*indiv_time_of_day_charts[3:]).resolve_scale(y ='shared')
).add_params(selection, dropdown_param).properties(
    title={
        "text": f"Sofi {MATCH_DAY_TYPE.title()} Matches on {', '.join(MATCH_DAYS)}", 
        "subtitle": [f"Match Starts: {MATCH_TOD}", "Travel Changes: am_peak, midday, pm_peak"]
}).resolve_scale(y ='shared')

MATCH_TOD = "pm_peak"
MATCH_DAY_TYPE = "weekday"

filtered_sofi_dict = event_helpers.grab_matches_by_day_type(
    wc_vars.sofi_match_times, MATCH_TOD, MATCH_DAY_TYPE)

MATCH_DAYS = list(filtered_sofi_dict.keys())

operator_dropdown = alt.binding_select(
    options=list_of_operators,
    name="GTFS Schedule: ",
)

dropdown_param = alt.selection_point(
    fields=["schedule_name"], value=list_of_operators[0], bind=operator_dropdown
)

selection = alt.selection_point(fields=["route_id_array"], bind="legend")

indiv_time_of_day_charts = [
    chart_utils.arrivals_by_time_of_day_grouped_bar(
        stop_tod_near2[stop_tod_near2.service_date.isin(["nonevent"] + MATCH_DAYS)], 
        time_of_day, 
        MATCH_DAY_TYPE
    ).properties(
        title = f"{time_of_day}", width=200, height=100
    ).add_params(selection, dropdown_param).transform_filter(dropdown_param) 
    for time_of_day in ["midday", "pm_peak", "evening", "early_am", "am_peak", "owl"]
    # set time-of-day order to put times surrounding match at the top
]    

alt.vconcat(
    alt.hconcat(*indiv_time_of_day_charts[:3]).resolve_scale(y ='shared'),
    alt.hconcat(*indiv_time_of_day_charts[3:]).resolve_scale(y ='shared')
).add_params(selection, dropdown_param).properties(
    title={
        "text": f"Sofi {MATCH_DAY_TYPE.title()} Matches on {', '.join(MATCH_DAYS)}", 
        "subtitle": [f"Match Starts: {MATCH_TOD}", "Travel Changes: midday, pm_peak, evening"]
}).resolve_scale(y ='shared')