def add_wait_time_stats_for_route(all_stats, timestamp_intervals, route_config, history_df):
route_id = route_config.id
history_df = history_df.sort_values('DEPARTURE_TIME', axis=0)
sid_values = history_df['SID'].values
did_values = history_df['DID'].values
for dir_info in route_config.get_direction_infos():
dir_id = dir_info.id
stop_ids = dir_info.get_stop_ids()
for i, stop_id in enumerate(stop_ids):
stop_df = history_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):
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:
dir_stats = all_stats['combined'][interval_index][route_id]['directions'][dir_id]
dir_stats['medianWaitTimes'][stop_id] = round(median_wait_time, 1)
for interval_index, (start_time, end_time) in enumerate(timestamp_intervals):
add_median_wait_time_stats_for_direction(
all_stats['combined'][interval_index][route_id]['directions'][dir_id]
)
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])
def compute_wait_times(agency_id,
d: date,
routes,
tz,
stat_ids,
save_to_s3=False):
print(d)
all_wait_time_stats = {}
time_str_intervals = constants.DEFAULT_TIME_STR_INTERVALS.copy()
time_str_intervals.append(('07:00', '19:00'))
timestamp_intervals = [
(int(util.get_localized_datetime(d, start_time_str, tz).timestamp()),
int(util.get_localized_datetime(d, end_time_str, tz).timestamp()))
for start_time_str, end_time_str in time_str_intervals
]
timestamp_intervals.append((None, None))
time_str_intervals.append((None, None))
for interval_index, _ in enumerate(timestamp_intervals):
all_wait_time_stats[interval_index] = {}
for stat_id in stat_ids:
all_wait_time_stats[interval_index][stat_id] = {}
for route in routes:
route_id = route.id
print(route_id)
route_config = nextbus.get_route_config(agency_id, route_id)
try:
history = arrival_history.get_by_date(agency_id, route_id, d)
except FileNotFoundError as ex:
print(ex)
continue
for interval_index, _ in enumerate(timestamp_intervals):
for stat_id in stat_ids:
all_wait_time_stats[interval_index][stat_id][route_id] = {}
df = history.get_data_frame()
df = df.sort_values('TIME', axis=0)
for dir_info in route_config.get_direction_infos():
dir_id = dir_info.id
for interval_index, _ in enumerate(timestamp_intervals):
for stat_id in stat_ids:
all_wait_time_stats[interval_index][stat_id][route_id][
dir_id] = {}
stop_ids = dir_info.get_stop_ids()
sid_values = df['SID'].values
for i, stop_id in enumerate(stop_ids):
stop_df = df[sid_values == stop_id]
all_time_values = stop_df['TIME'].values
for interval_index, (
start_time,
end_time) in enumerate(timestamp_intervals):
wait_time_stats = wait_times.get_stats(
all_time_values, start_time, end_time)
add_wait_time_stats_for_stop(
all_wait_time_stats[interval_index], stat_ids,
route_id, dir_id, stop_id, wait_time_stats)
for interval_index, _ in enumerate(timestamp_intervals):
for stat_id in stat_ids:
add_median_wait_time_stats_for_direction(
all_wait_time_stats[interval_index][stat_id][route_id]
[dir_id], stat_id)
for interval_index, (start_time,
end_time) in enumerate(timestamp_intervals):
start_time_str, end_time_str = time_str_intervals[interval_index]
for stat_id in stat_ids:
stat = stat_groups[stat_id]
data_str = json.dumps(
{
'version': wait_times.DefaultVersion,
'start_time': start_time,
'end_time': end_time,
'stat': stat,
'routes': all_wait_time_stats[interval_index][stat_id]
},
separators=(',', ':'))
cache_path = wait_times.get_cache_path(agency_id, d, stat_id,
start_time_str,
end_time_str)
cache_dir = Path(cache_path).parent
if not cache_dir.exists():
cache_dir.mkdir(parents=True, exist_ok=True)
print(f'saving to {cache_path}')
with open(cache_path, "w") as f:
f.write(data_str)
if save_to_s3:
s3 = boto3.resource('s3')
s3_path = wait_times.get_s3_path(agency_id, d, stat_id,
start_time_str, end_time_str)
s3_bucket = wait_times.get_s3_bucket()
print(f'saving to s3://{s3_bucket}/{s3_path}')
object = s3.Object(s3_bucket, s3_path)
object.put(Body=gzip.compress(bytes(data_str, 'utf-8')),
CacheControl='max-age=86400',
ContentType='application/json',
ContentEncoding='gzip',
ACL='public-read')
hist = arrival_history.get_by_date(agency, route_id, d)
arrivals = hist.get_data_frame(stop_id=stop, direction_id=stop_dir)
start_time = util.get_timestamp_or_none(d, start_time_str, tz)
end_time = util.get_timestamp_or_none(d, end_time_str, tz)
departure_times = np.sort(arrivals['DEPARTURE_TIME'].values)
if len(departure_times) == 0:
continue
first_bus_date_times.append(
datetime.fromtimestamp(departure_times[0], tz))
last_bus_date_times.append(
datetime.fromtimestamp(departure_times[-1], tz))
stats = wait_times.get_stats(departure_times, start_time, end_time)
average = stats.get_average()
if average is not None:
averages.append(average)
quantiles = stats.get_quantiles([0, 0.1, 0.5, 0.9, 1])
if quantiles is not None:
minimums.append(quantiles[0])
p10s.append(quantiles[1])
medians.append(quantiles[2])
p90s.append(quantiles[3])
maximums.append(quantiles[4])
print(f"Date: {', '.join([str(date) for date in dates])}")
print(f"Local Time Range: [{start_time_str}, {end_time_str}]")
print(f"Route: {route_id} ({route_config.title})")