def get_dropoffs(lon, lat):
# For the approximated point get the dropoff points from the dropoffs table
curr.execute("select drops from dropoffs_10 where lon = {0} and lat = {1}".format(lon, lat))
result = curr.fetchall()
if len(result) > 0:
return result[0][0]
else:
return []
def get_nearest_point(lon, lat):
# Get the nearest discretized point (this will be the user destination going forward)
curr.execute("select lon, lat, ST_DistanceSphere(ST_MakePoint(lon, lat), ST_MakePoint(" + str(lon) + "," + str(lat) + ")) as dist " +
"from disc_manhattan_points where ST_DWithin(ST_SetSRID(ST_MakePoint(lon, lat), 4326), ST_SetSRID(ST_MakePoint(" + str(lon) + "," + str(lat) + "), 4326), 3000)" +
"order by dist limit 1")
point = curr.fetchone()
approximated_lon = point[0]
approximated_lat = point[1]
dist = point[2]
return [approximated_lon, approximated_lat, dist]
def check_walk_shareable(src, dest1, delay1, drops1, dest2, delay2, drops2,
src_dest1_dist, src_dest2_dist):
curr.execute("select dist, time from lag_pt_dist_time where lon1 = " +
str(src[0]) + " and lat1 = " + str(src[1]) + " and lon2 = " +
str(dest1[0]) + " and lat2 = " + str(dest1[1]))
result = curr.fetchall()
# src_dest1_dist = float(result[0])
if len(result) > 0:
src_dest1_time = float(result[0][1] / 60)
else:
src_dest1_time = ((src_dest1_dist / 35) * 60) + 1.6
curr.execute("select dist, time from lag_pt_dist_time where lon1 = " +
str(src[0]) + " and lat1 = " + str(src[1]) + " and lon2 = " +
str(dest2[0]) + " and lat2 = " + str(dest2[1]))
result = curr.fetchall()
# src_dest2_dist = float(result[0])
if len(result) > 0:
src_dest2_time = float(result[0][1] / 60)
else:
src_dest2_time = ((src_dest2_dist / 35) * 60) + 1.6
for drop in drops1:
# Drop rider1 first
src_drop_dist = get_distance(src[0], src[1], drop[0], drop[1])
src_drop_time = ((src_drop_dist / 35) * 60) + 1.6
drop_dest2_dist = get_distance(drop[0], drop[1], dest2[0], dest2[1])
drop_dest2_time = ((drop_dest2_dist / 35) * 60) + 1.6
if (src_drop_dist + drop_dest2_dist) < (src_dest1_dist + src_dest2_dist) and \
(src_drop_time + drop_dest2_time) <= (src_dest2_time + delay2):
return (src_dest1_dist + src_dest2_dist) - (src_drop_dist +
drop_dest2_dist)
for drop in drops2:
# Drop rider2 first
src_drop_dist = get_distance(src[0], src[1], drop[0], drop[1])
src_drop_time = ((src_drop_dist / 35) * 60) + 1.6 # Error of ~ 1.6 min
drop_dest1_dist = get_distance(drop[0], drop[1], dest1[0], dest1[1])
drop_dest1_time = ((src_drop_dist / 35) * 60) + 1.6
if (src_drop_dist + drop_dest1_dist) < (src_dest1_dist + src_dest2_dist) and \
(src_drop_time + drop_dest1_time) <= (src_dest1_time + delay1):
return (src_dest1_dist + src_dest2_dist) - (src_drop_dist +
drop_dest1_dist)
return -1
def no_walks_shareable(dest1, delay1, dest2, delay2, src, src_dest1_dist,
src_dest2_dist):
dest1_dest2_dist = get_distance(dest1[0], dest1[1], dest2[0], dest2[1])
dest2_dest1_dist = dest1_dest2_dist
dest1_dest2_time = ((dest1_dest2_dist / 35) * 60) + 1.6
dest2_dest1_time = dest1_dest2_time
curr.execute("select time from lag_pt_dist_time where lon1 = " +
str(src[0]) + " and lat1 = " + str(src[1]) + " and lon2 = " +
str(dest1[0]) + " and lat2 = " + str(dest1[1]))
result = curr.fetchall()
if len(result) > 0:
src_dest1_time = float(result[0][0] / 60)
else:
src_dest1_time = ((src_dest1_dist / 35) * 60) + 1.6
curr.execute("select time from lag_pt_dist_time where lon1 = " +
str(src[0]) + " and lat1 = " + str(src[1]) + " and lon2 = " +
str(dest2[0]) + " and lat2 = " + str(dest2[1]))
result = curr.fetchall()
if len(result) > 0:
src_dest2_time = float(result[0][0] / 60)
else:
src_dest2_time = ((src_dest2_dist / 35) * 60) + 1.6
if (src_dest1_dist + dest1_dest2_dist) < (src_dest1_dist + src_dest2_dist) and \
(src_dest1_time + dest1_dest2_time) <= (src_dest2_time + delay2):
return (src_dest1_dist + src_dest2_dist) - (src_dest1_dist +
dest1_dest2_dist)
elif (src_dest2_dist + dest2_dest1_dist) < (src_dest1_dist + src_dest2_dist) and \
(src_dest2_time + dest2_dest1_time) <= (src_dest1_time + delay1):
return (src_dest1_dist + src_dest2_dist) - (src_dest2_dist +
dest2_dest1_dist)
else:
return -1
def get_distance(lon1, lat1, lon2, lat2):
curr.execute("select ST_DistanceSphere(ST_MakePoint(" + str(lon1) + "," + str(lat1) + "), ST_MakePoint(" + str(lon2) +
"," + str(lat2) + "))")
distance = curr.fetchall()
return (float(distance[0][0]) + 3000) / 1600
def main_from(pool, pno):
pool_info = {}
start_time = timeit.default_timer()
pool_info[pno] = {}
dist = {}
individual_trip_dist = {}
total_num_trips = len(pool[pno]["destination"])
total_dist = sum(pool[pno]["distance"])
distance_error = 0
pool_info[pno]["total_num_trips"] = total_num_trips
pool_info[pno]["total_dist"] = total_dist
shareable_trips = []
# Go over each trip A
for i in range(len(pool[pno]["destination"])):
min_non_share_dist = sys.maxsize
if i not in individual_trip_dist.keys():
individual_trip_dist[i] = pool[pno]["distance"][i]
num_shareable = 0
print("Trip {0} started".format(i + 1))
dist[i + 1] = {}
trip_1 = pool[pno]["destination"][i]
dest1 = get_nearest_point(trip_1[0], trip_1[1])
distance_error += dest1[2]
drops1 = get_dropoffs(dest1[0], dest1[1])
curr.execute("select time from lag_pt_dist_time where lon2 = " +
str(dest1[0]) + " and lat2 = " + str(dest1[1]))
result = curr.fetchall()
if len(result) > 0:
delay1 = 0.2 * float(result[0][0] / 60) # 20% delay
else: # No travel time is precomputed for this pair of points
travel_time = ((individual_trip_dist[i] / 35) * 60) + 1.6
delay1 = 0.2 * float(travel_time)
for j in range(i + 1, len(
pool[pno]["destination"])): # Go over every other trip B
if j not in individual_trip_dist.keys():
individual_trip_dist[j] = pool[pno]["distance"][j]
trip_2 = pool[pno]["destination"][j]
dest2 = get_nearest_point(trip_2[0], trip_2[1])[:-1]
new_distance = get_distance(dest1[0], dest1[1], dest2[0], dest2[1])
# To remove optimization and process all combination of trips, comment the following 2 lines
if new_distance > min_non_share_dist:
continue
drops2 = get_dropoffs(dest2[0], dest2[1])
curr.execute("select time from lag_pt_dist_time where lon2 = " +
str(dest2[0]) + " and lat2 = " + str(dest2[1]))
result = curr.fetchall()
if len(result) > 0:
delay2 = 0.2 * float(result[0][0] / 60)
else: # No travel time precomputed for this pair of points
travel_time = ((individual_trip_dist[i] / 35) * 60) + 1.6
delay2 = 0.2 * float(travel_time)
dist_saved = no_walks_shareable(dest1, delay1, dest2, delay2,
[-73.874, 40.774],
individual_trip_dist[i],
individual_trip_dist[j])
if dist_saved != -1:
num_shareable += 1
shareable_trips.append((i + 1, j + 1, round(dist_saved)))
dist[i + 1][j + 1] = dist_saved
if (j + 1) not in dist.keys():
dist[j + 1] = {}
dist[j + 1][i + 1] = dist_saved
else:
dist_saved = check_walk_shareable([-73.874, 40.774], dest1,
delay1, drops1, dest2,
delay2, drops2,
individual_trip_dist[i],
individual_trip_dist[j])
if dist_saved != -1:
num_shareable += 1
shareable_trips.append((i + 1, j + 1, round(dist_saved)))
dist[i + 1][j + 1] = dist_saved
if (j + 1) not in dist.keys():
dist[j + 1] = {}
dist[j + 1][i + 1] = dist_saved
else:
# Update min_non_share_distance
if new_distance < min_non_share_dist:
min_non_share_dist = new_distance
print("Trip {0} done. {1} trips shareable with this trip.".format(
i + 1, num_shareable))
distance_error /= pool_info[pno]["total_num_trips"]
pool_info[pno]["average_distance_error"] = distance_error
pool_info = max_match(shareable_trips, pno, pool_info, dist,
timeit.default_timer() - start_time)
print("Pool {0} done.".format(pno))
return pool_info
def main_to(pool, pno):
pool_info = {}
individual_trip_dist = {}
start_time = timeit.default_timer()
distance_error = 0
dist = {}
pool_info[pno] = {}
total_num_trips = len(pool[pno]["destination"])
total_dist = sum(pool[pno]["distance"])
pool_info[pno]["total_num_trips"] = total_num_trips
pool_info[pno]["total_dist"] = total_dist
shareable_trips = []
for i in range(len(pool[pno]["pickup"])):
min_non_share_dist = sys.maxsize
if i not in individual_trip_dist.keys():
individual_trip_dist[i] = pool[pno]["distance"][i]
num_shareable = 0
dist[i + 1] = {}
trip_1 = pool[pno]["pickup"][i]
src1 = get_nearest_point(trip_1[0], trip_1[1])
distance_error += src1[2]
curr.execute("select time from lag_pt_dist_time where lon2 = " +
str(src1[0]) + " and lat2 = " + str(src1[1]))
result = curr.fetchall()
if len(result) > 0:
delay1 = 0.2 * float(result[0][0] / 60)
else:
travel_time = ((individual_trip_dist[i] / 35) * 60) + 1.6
delay1 = 0.2 * float(travel_time)
for j in range(i + 1, len(pool[pno]["pickup"])):
if j not in individual_trip_dist.keys():
individual_trip_dist[j] = pool[pno]["distance"][j]
trip_2 = pool[pno]["pickup"][j]
src2 = get_nearest_point(trip_2[0], trip_2[1])[:-1]
new_distance = get_distance(src1[0], src1[1], src2[0], src2[1])
# To remove optimization and process all combination of trips, comment the following 2 lines
if new_distance > min_non_share_dist:
continue
curr.execute("select time from lag_pt_dist_time where lon2 = " +
str(src2[0]) + " and lat2 = " + str(src2[1]))
result = curr.fetchall()
if len(result) > 0:
delay2 = 0.2 * float(result[0][0] / 60)
else:
travel_time = ((individual_trip_dist[i] / 35) * 60) + 1.6
delay2 = 0.2 * float(travel_time)
dist_saved = check_shareable_to_laguardia(src1, delay1, src2,
delay2,
individual_trip_dist[i],
individual_trip_dist[j])
if dist_saved != -1:
num_shareable += 1
shareable_trips.append((i + 1, j + 1, round(dist_saved)))
dist[i + 1][j + 1] = dist_saved
if (j + 1) not in dist.keys():
dist[j + 1] = {}
dist[j + 1][i + 1] = dist_saved
else:
# Update min_non_share_dist
min_non_share_dist = new_distance
print("Trip {0} done. {1} trips shareable with this trip.".format(
i + 1, num_shareable))
distance_error /= pool_info[pno]["total_num_trips"]
pool_info[pno]["average_distance_error"] = distance_error
pool_info = max_match(shareable_trips, pno, pool_info, dist,
timeit.default_timer() - start_time)
print("Pool {0} done.\n".format(pno))
return pool_info
