def add_headway_and_wait_time_stats_for_route(all_stats, timestamp_intervals,
route_config, df):
route_id = route_config.id
df = df.sort_values('DEPARTURE_TIME', axis=0)
sid_values = df['SID'].values
did_values = df['DID'].values
for dir_info in route_config.get_direction_infos():
dir_id = dir_info.id
stop_ids = dir_info.get_stop_ids()
all_median_headways = collections.defaultdict(list)
all_median_wait_times = collections.defaultdict(list)
for i, stop_id in enumerate(stop_ids):
stop_df = df[(sid_values == stop_id) & (did_values == dir_id)]
all_time_values = stop_df['DEPARTURE_TIME'].values
for interval_index, (start_time,
end_time) in enumerate(timestamp_intervals):
dir_stats = all_stats[StatIds.Combined][interval_index][
route_id]['directions'][dir_id]
headways = metrics.compute_headway_minutes(
all_time_values, start_time, end_time)
if len(headways) > 0:
all_median_headways[interval_index].append(
np.median(headways))
wait_time_stats = wait_times.get_stats(all_time_values,
start_time, end_time)
median_wait_time = wait_time_stats.get_quantile(0.5)
if median_wait_time is not None:
all_median_wait_times[interval_index].append(
median_wait_time)
all_stats[StatIds.MedianTripTimes][interval_index][
route_id]['directions'][dir_id]['medianWaitTimes'][
stop_id] = round(median_wait_time, 1)
for interval_index, (start_time,
end_time) in enumerate(timestamp_intervals):
dir_stats = all_stats[StatIds.Combined][interval_index][route_id][
'directions'][dir_id]
dir_stats['medianWaitTime'] = get_median_or_none(
all_median_wait_times[interval_index])
dir_stats['medianHeadway'] = get_median_or_none(
all_median_headways[interval_index])
get_history_ms += t2 - t1
get_data_frame_ms += t3 - t2
df = pd.concat(route_dfs)
if df.empty:
print(f"no arrival times found for stop {stop_id} on {date_str}")
continue
df = df.sort_values('TIME', axis=0)
t4 = time.time() * 1000
df['headway_min'] = np.r_[
np.nan, metrics.compute_headway_minutes(df['TIME'].values)]
df['DATE_TIME'] = df.TIME.apply(
lambda t: datetime.fromtimestamp(t, tz))
t5 = time.time() * 1000
for row in df.itertuples():
did = row.DID
dir_info = [
dir_info for dir_info in dir_infos if dir_info.id == did
][0]
dist_str = f'{row.DIST}'.rjust(3)
dwell_time = util.render_dwell_time(row.DEPARTURE_TIME - row.TIME)
headway_str = f'{round(row.headway_min, 1)}'.rjust(4)
print(
f"{row.DATE_TIME.date()} {row.DATE_TIME.time()} ({row.TIME}) {dwell_time} vid:{row.VID} {dist_str}m {headway_str} min ({row.ROUTE} - {dir_info.title})"
timetable = timetables.get_by_date(agency_id, route_id, d)
route_config = agency.get_route_config(route_id)
tz = agency.tz
direction_id = args.dir
timetable_df = timetable.get_data_frame(
stop_id=stop_id, direction_id=direction_id).sort_values('TIME')
early_sec = early_min * 60
late_sec = late_min * 60
timetable_df['scheduled_headway'] = np.r_[
np.nan,
metrics.compute_headway_minutes(timetable_df['TIME'].values)]
if comparison:
history = arrival_history.get_by_date(agency_id, route_id, d)
arrivals_df = history.get_data_frame(stop_id=stop_id,
direction_id=direction_id)
comparison_df = timetables.match_schedule_to_actual_times(
timetable_df['TIME'].values,
arrivals_df['TIME'].values,
early_sec=early_sec,
late_sec=late_sec)
timetable_df = pd.concat([timetable_df, comparison_df], axis=1)
timetable_df['DATE_TIME'] = timetable_df['TIME'].apply(
route_dfs.append(route_df)
get_history_ms += t2 - t1
get_data_frame_ms += t3 - t2
df = pd.concat(route_dfs)
if df.empty:
print(f"no arrival times found for stop {stop_id} on {date_str}")
continue
df = df.sort_values('TIME', axis=0)
t4 = time.time() * 1000
df['headway_min'] = metrics.compute_headway_minutes(df)
t5 = time.time() * 1000
for index, row in df.iterrows():
did = row.DID
dir_info = [
dir_info for dir_info in dir_infos if dir_info.id == did
][0]
print(
f"t={row.DATE_STR} {row.TIME_STR} ({row.TIME}) v:{row.VID} {round(row.headway_min,1)} min ({row.ROUTE} - {dir_info.title})"
)
t6 = time.time() * 1000
day_headways = df.headway_min[df.headway_min.notnull()]
def metrics_page():
metrics_start = time.time()
route_id = request.args.get('route_id')
if route_id is None:
route_id = '12'
start_stop_id = request.args.get('start_stop_id')
if start_stop_id is None:
start_stop_id = '3476'
end_stop_id = request.args.get('end_stop_id')
direction_id = request.args.get('direction_id')
start_date_str = request.args.get('start_date')
end_date_str = request.args.get('end_date')
date_str = request.args.get('date')
if date_str is not None:
start_date_str = end_date_str = date_str
else:
if start_date_str is None:
start_date_str = '2019-02-01'
if end_date_str is None:
end_date_str = start_date_str
start_time_str = request.args.get(
'start_time') # e.g. "14:00" (24h time of day)
end_time_str = request.args.get(
'end_time') # e.g. "18:00" (24h time of day)
params = {
'start_stop_id': start_stop_id,
'end_stop_id': end_stop_id,
'route_id': route_id,
'direction_id': direction_id,
'start_date': start_date_str,
'end_date': end_date_str,
'start_time': start_time_str,
'end_time': end_time_str,
}
try:
dates = util.get_dates_in_range(start_date_str, end_date_str)
except Exception as ex:
return Response(json.dumps({
'params': params,
'error': str(ex),
},
indent=2),
status=400,
mimetype='application/json')
tz = pytz.timezone('US/Pacific')
route_config = nextbus.get_route_config('sf-muni', route_id)
start_stop_info = route_config.get_stop_info(start_stop_id)
end_stop_info = route_config.get_stop_info(
end_stop_id) if end_stop_id else None
# 404 if the given stop isn't on the route
# TODO: what should be done for the case where the start stop id is valid but the end stop id isn't?
if start_stop_info is None:
return Response(json.dumps(
{
'params': params,
'error': f"Stop {start_stop_id} is not on route {route_id}",
},
indent=2),
status=404,
mimetype='application/json')
if direction_id is not None:
dir_info = route_config.get_direction_info(direction_id)
if dir_info is not None:
dir_infos = [dir_info]
else:
dir_infos = []
else:
# TODO: validation for end_stop_id directions if given (see trips.py)
dirs = route_config.get_directions_for_stop(start_stop_id)
dir_infos = [
route_config.get_direction_info(direction) for direction in dirs
]
if end_stop_id:
end_stop_dirs = route_config.get_directions_for_stop(end_stop_id)
both_stops_same_dir = direction_id in end_stop_dirs
directions = [{
'id': dir_info.id,
'title': dir_info.title
} for dir_info in dir_infos]
headway_min_arr = []
waits = []
if end_stop_id:
completed_trips = []
for d in dates:
try:
history = arrival_history.get_by_date('sf-muni', route_id, d)
df = history.get_data_frame(start_stop_id,
tz=tz,
direction_id=direction_id,
start_time_str=start_time_str,
end_time_str=end_time_str)
# get all headways for the selected stop (arrival time minus previous arrival time), computed separately for each day
df['headway_min'] = metrics.compute_headway_minutes(df)
# temporarily skip calculation of wait times until data is shown in front end
waits.append(
wait_times.get_waits(df, start_stop_info, d, tz, route_id,
start_time_str, end_time_str))
if end_stop_id and both_stops_same_dir:
trips = trip_times.get_trip_times(df, history, tz,
start_stop_id, end_stop_id)
completed_trips.append(
trips.trip_min[trips.trip_min.notnull()])
headway_min = df.headway_min[df.headway_min.notnull(
)] # remove NaN row (first bus of the day)
headway_min_arr.append(df.headway_min)
except FileNotFoundError as ex:
return Response(json.dumps(
{
'params':
params,
'error':
f"Arrival history not found for route {route_id} on {d.isoformat()}",
},
indent=2),
status=404,
mimetype='application/json')
except IndexError as ex:
return Response(json.dumps(
{
'params':
params,
'error':
f"No arrivals found for stop {start_stop_id} on route {route_id} in direction {direction_id} on {d.isoformat()}",
},
indent=2),
status=404,
mimetype='application/json')
headway_min = pd.concat(headway_min_arr)
waits = pd.concat(waits)
if end_stop_id and both_stops_same_dir:
completed_trips = pd.concat(completed_trips)
if headway_min.empty:
return Response(json.dumps(
{
'params':
params,
'error':
f"No arrivals for stop {start_stop_id} on route {route_id}",
},
indent=2),
status=404,
mimetype='application/json')
data = {
'params':
params,
'route_title':
route_config.title,
'start_stop_title':
start_stop_info.title if start_stop_info else None,
'end_stop_title':
end_stop_info.title if end_stop_info else None,
'directions':
directions,
'headway_min':
metrics.get_headways_stats(headway_min),
'wait_times':
metrics.get_wait_times_stats(waits, tz),
'trip_times':
metrics.get_trip_times_stats(completed_trips, start_stop_id,
end_stop_id)
if end_stop_id and both_stops_same_dir else None,
}
metrics_end = time.time()
data['processing_time'] = (metrics_end - metrics_start)
return Response(json.dumps(data, indent=2), mimetype='application/json')