diff options
Diffstat (limited to 'utils.py')
| -rw-r--r-- | utils.py | 72 |
1 files changed, 50 insertions, 22 deletions
@@ -16,18 +16,20 @@ def cluster_and_optimize(df, centroids, end, time_diff=0.25, max_time=24, n=2): :return: a dataframe with a cluster column """ # Create a new column with normalized gps coordinates and centroids - df['normalized_gps'], norm_centroids = __normalize_gps(df['gps'].values.tolist(), centroids) + df["normalized_gps"], norm_centroids = __normalize_gps( + df["gps"].values.tolist(), centroids + ) # Cluster the coordinates kmeans = KMeans(n_clusters=len(norm_centroids), init=norm_centroids) - kmeans.fit(df['normalized_gps'].values.tolist()) + kmeans.fit(df["normalized_gps"].values.tolist()) - df['cluster'] = kmeans.labels_ + df["cluster"] = kmeans.labels_ routes = [] starts = [] for i in range(len(centroids)): - routes.append(df[df['cluster'] == i]['gps'].values.tolist()) + routes.append(df[df["cluster"] == i]["gps"].values.tolist()) starts.append(list_to_string([centroids[i]])) routes = __minimize_route_time_diff(routes, starts, end, time_diff, n) @@ -35,7 +37,7 @@ def cluster_and_optimize(df, centroids, end, time_diff=0.25, max_time=24, n=2): # Remove waypoints from the longest route until the trip time is less than the max time for i in range(len(routes)): routes[i] = __remove_longest_waypoints(routes[i], starts[i], end, max_time) - df.loc[df['gps'].astype(str).isin(map(str, routes[i])), 'cluster'] = i + df.loc[df["gps"].astype(str).isin(map(str, routes[i])), "cluster"] = i return df, routes @@ -46,9 +48,9 @@ def list_to_string(list_of_lists): :param list_of_lists: a list of lists :return: a string of the list of lists """ - string = '' + string = "" for i in list_of_lists: - string += str(i[1]) + ',' + str(i[0]) + ';' + string += str(i[1]) + "," + str(i[0]) + ";" return string @@ -62,20 +64,25 @@ def create_json_df(coordinate_string, start, end): :return: a dataframe of the coordinates in order of the waypoint index """ coordinates = requests.get( - 'http://acetyl.net:5000/trip/v1/bike/' + start + coordinate_string + end + '?roundtrip=false&source=first&destination=last') + "http://acetyl.net:5000/trip/v1/bike/" + + start + + coordinate_string + + end + + "?roundtrip=false&source=first&destination=last" + ) coordinates = coordinates.json() # Create a dataframe from the JSON - df = pd.DataFrame(coordinates['waypoints']) + df = pd.DataFrame(coordinates["waypoints"]) # Separate the location column into lon and lat columns - df['lat'] = df['location'].apply(lambda x: x[0]) - df['lon'] = df['location'].apply(lambda x: x[1]) + df["lat"] = df["location"].apply(lambda x: x[0]) + df["lon"] = df["location"].apply(lambda x: x[1]) - df['waypoint_index'] = df['waypoint_index'].astype(int) + df["waypoint_index"] = df["waypoint_index"].astype(int) # Map out the waypoints in order of the waypoint index - df = df.sort_values(by=['waypoint_index']) + df = df.sort_values(by=["waypoint_index"]) return df @@ -91,10 +98,15 @@ def get_trip_time(coordinate_string, num_waypoints, start, end, time_per_waypoin :return: the trip time in hours """ coordinates = requests.get( - 'http://acetyl.net:5000/trip/v1/bike/' + start + coordinate_string + end + '?roundtrip=false&source=first&destination=last') + "http://acetyl.net:5000/trip/v1/bike/" + + start + + coordinate_string + + end + + "?roundtrip=false&source=first&destination=last" + ) coordinates = coordinates.json() - travel_time_seconds = int(coordinates['trips'][0]['duration']) + travel_time_seconds = int(coordinates["trips"][0]["duration"]) waypoint_time_seconds = num_waypoints * time_per_waypoint total_time_hours = (travel_time_seconds + waypoint_time_seconds) / 3600 @@ -129,8 +141,9 @@ def __minimize_route_time_diff(routes, starts, end, time_diff, n): # If the difference is greater than the time difference, move a coordinate from the longest route to the shortest route if time_difference > time_diff: # Move a coordinate from the longest route to the shortest route - closest_coordinate = __find_closest_coordinate(routes[sorted_indices[-1]], - __mean_center(routes[sorted_indices[0]])) + closest_coordinate = __find_closest_coordinate( + routes[sorted_indices[-1]], __mean_center(routes[sorted_indices[0]]) + ) routes[sorted_indices[0]].append(closest_coordinate) routes[sorted_indices[-1]].remove(closest_coordinate) @@ -151,7 +164,9 @@ def __remove_longest_waypoints(route_coordinates, start, end, max_time): :return: a list of coordinates """ # Find the trip time for the route - route_time = get_trip_time(list_to_string(route_coordinates), len(route_coordinates), start, end) + route_time = get_trip_time( + list_to_string(route_coordinates), len(route_coordinates), start, end + ) # If the trip time is greater than the max time, remove the waypoint with the longest distance from the mean if route_time > max_time: @@ -189,10 +204,18 @@ def __normalize_gps(coordinates, centroids): for i in coordinates: normalized_coordinates.append( - [__min_max_normalize(i[0], min_lat, max_lat), __min_max_normalize(i[1], min_lon, max_lon)]) + [ + __min_max_normalize(i[0], min_lat, max_lat), + __min_max_normalize(i[1], min_lon, max_lon), + ] + ) for i in centroids: normalized_centroids.append( - [__min_max_normalize(i[0], min_lat, max_lat), __min_max_normalize(i[1], min_lon, max_lon)]) + [ + __min_max_normalize(i[0], min_lat, max_lat), + __min_max_normalize(i[1], min_lon, max_lon), + ] + ) return normalized_coordinates, normalized_centroids @@ -250,7 +273,10 @@ def __mean_center(coordinates): :param coordinates: the list of coordinates :return: the mean center of the coordinates """ - return [sum([i[0] for i in coordinates]) / len(coordinates), sum([i[1] for i in coordinates]) / len(coordinates)] + return [ + sum([i[0] for i in coordinates]) / len(coordinates), + sum([i[1] for i in coordinates]) / len(coordinates), + ] def __distance(coordinate1, coordinate2): @@ -260,4 +286,6 @@ def __distance(coordinate1, coordinate2): :param coordinate2: the second coordinate :return: the distance between the two coordinates """ - return ((coordinate1[0] - coordinate2[0]) ** 2 + (coordinate1[1] - coordinate2[1]) ** 2) ** 0.5 + return ( + (coordinate1[0] - coordinate2[0]) ** 2 + (coordinate1[1] - coordinate2[1]) ** 2 + ) ** 0.5 |