def choose(situation, target, distance, mode, stationFirst, nbCandidates=5):
"""
Choose the best station around the arrival and then define a trip
between the departure and the station.
"""
city = situation["city"]
people = situation["people"]
time = tb.convert_time(situation["time"])
# Will convert the positions to (lat, lon) if they are textual addresses
departure = geography.convert_to_point(city, situation["departure"])
arrival = geography.convert_to_point(city, situation["arrival"])
# Find the close stations with MongoDB
stations = [station for station in query.close_points(city, arrival, number=nbCandidates)]
# Get the distances to the stations
candidates = geography.compute_distances_manual(arrival, stations, distance)
# Sort the stations by distance
candidates.sort(key=lambda station: station["duration"])
# Find an appropriate solution through the sorted candidates
trip = False
for candidate in candidates:
# Calculate what time it would be when reaching the candidate station
forecast = tb.epoch_to_datetime(time + candidate["duration"])
# Extract features from the forecasted time
variables = munging.temporal_features(forecast)
features = [variables["hour"], variables["minute"], variables["weekday"]]
# Check if the prediction is satisfying
settings = tb.read_json("common/settings.json")
method = eval(settings["learning"]["method"])
# Make a prediction
prediction = method.predict(features, target, city, candidate["_id"])
bias = tb.read_json("common/settings.json")["learning"]["bias"]
if target == "bikes":
bias *= -1
if prediction + bias >= people:
stationPosition = list(reversed(candidate["p"]))
if stationFirst is False:
trip = tb.reshape(mode, departure, stationPosition)
else:
trip = tb.reshape(mode, stationPosition, departure)
break
return trip
def full_trip(situation):
''' We can use the previous functions. '''
# Find the route from the departure to the departure station
situationOne = deepcopy(situation)
situationOne['arrival'] = situationOne['departure']
firstPath = choose(situationOne, 'bikes', 'walking', 'walking', False)
# Find the route from the arrival station to the arrival
situationTwo = deepcopy(situation)
situationTwo['departure'] = situationTwo['arrival']
secondPath = choose(situationTwo, 'spaces', 'walking', 'walking', True)
# Find the route between both stations
firstStation = firstPath['points'][1]
secondStation = secondPath['points'][0]
A = [firstStation['lat'], firstStation['lon']]
B = [secondStation['lat'], secondStation['lon']]
intermediatePath = tb.reshape('bicycling', A, B)
# Generate the routes
routes = [
turn_by_turn(firstPath),
turn_by_turn(intermediatePath),
turn_by_turn(secondPath)
]
return routes
