Spaces:
Sleeping
Sleeping
Update app.py
Browse filesUpdated the app with more examples
app.py
CHANGED
@@ -1,3 +1,211 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
#!/usr/bin/env python
|
2 |
"""
|
3 |
Gradio App for NYC Taxi Fare Prediction & Road Route Visualization using OSRM
|
@@ -179,6 +387,28 @@ def predict_fare_and_map(plat, plong, dlat, dlong, psngr, dt):
|
|
179 |
f"Route Distance (OSRM): {route_distance_km:.2f} km")
|
180 |
return output_text, map_html
|
181 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
182 |
# -----------------------------
|
183 |
# Gradio Interface
|
184 |
# -----------------------------
|
@@ -196,12 +426,15 @@ iface = gr.Interface(
|
|
196 |
gr.Textbox(label="Prediction & Distance"),
|
197 |
gr.HTML(label="Map")
|
198 |
],
|
|
|
199 |
title="NYC Taxi Fare Prediction with OSRM Road Route",
|
200 |
description=(
|
201 |
"Enter pickup/dropoff coordinates, passenger count, and pickup datetime to predict the taxi fare. "
|
202 |
-
"The app displays the actual road route (blue line) from OSRM on a Folium map."
|
|
|
203 |
)
|
204 |
)
|
205 |
|
206 |
if __name__ == "__main__":
|
207 |
iface.launch()
|
|
|
|
1 |
+
# #!/usr/bin/env python
|
2 |
+
# """
|
3 |
+
# Gradio App for NYC Taxi Fare Prediction & Road Route Visualization using OSRM
|
4 |
+
|
5 |
+
# Requirements:
|
6 |
+
# pip install torch gradio requests polyline folium pandas numpy
|
7 |
+
# """
|
8 |
+
|
9 |
+
# import torch
|
10 |
+
# import torch.nn as nn
|
11 |
+
# import numpy as np
|
12 |
+
# import pandas as pd
|
13 |
+
# import requests
|
14 |
+
# import polyline
|
15 |
+
# import folium
|
16 |
+
# import gradio as gr
|
17 |
+
|
18 |
+
# # -----------------------------
|
19 |
+
# # Model Definition (TabularModel)
|
20 |
+
# # -----------------------------
|
21 |
+
# class TabularModel(nn.Module):
|
22 |
+
# def __init__(self, emb_szs, n_cont, out_sz, layers, p=0.5):
|
23 |
+
# """
|
24 |
+
# Model for tabular data combining embeddings for categorical variables and
|
25 |
+
# a feed-forward network for continuous features.
|
26 |
+
# """
|
27 |
+
# super().__init__()
|
28 |
+
# self.embeds = nn.ModuleList([nn.Embedding(ni, nf) for ni, nf in emb_szs])
|
29 |
+
# self.emb_drop = nn.Dropout(p)
|
30 |
+
# self.bn_cont = nn.BatchNorm1d(n_cont)
|
31 |
+
|
32 |
+
# n_emb = sum([nf for _, nf in emb_szs])
|
33 |
+
# n_in = n_emb + n_cont
|
34 |
+
|
35 |
+
# layerlist = []
|
36 |
+
# for i in layers:
|
37 |
+
# layerlist.append(nn.Linear(n_in, i))
|
38 |
+
# layerlist.append(nn.ReLU(inplace=True))
|
39 |
+
# layerlist.append(nn.BatchNorm1d(i))
|
40 |
+
# layerlist.append(nn.Dropout(p))
|
41 |
+
# n_in = i
|
42 |
+
# layerlist.append(nn.Linear(layers[-1], out_sz))
|
43 |
+
# self.layers = nn.Sequential(*layerlist)
|
44 |
+
|
45 |
+
# def forward(self, x_cat, x_cont):
|
46 |
+
# embeddings = []
|
47 |
+
# for i, e in enumerate(self.embeds):
|
48 |
+
# embeddings.append(e(x_cat[:, i]))
|
49 |
+
# x = torch.cat(embeddings, 1)
|
50 |
+
# x = self.emb_drop(x)
|
51 |
+
|
52 |
+
# x_cont = self.bn_cont(x_cont)
|
53 |
+
# x = torch.cat([x, x_cont], 1)
|
54 |
+
# x = self.layers(x)
|
55 |
+
# return x
|
56 |
+
|
57 |
+
# # -----------------------------
|
58 |
+
# # Load the trained model
|
59 |
+
# # -----------------------------
|
60 |
+
# # These parameters must match those used during training.
|
61 |
+
# emb_szs = [(24, 12), (2, 1), (7, 4)]
|
62 |
+
# n_cont = 6 # [pickup_lat, pickup_long, dropoff_lat, dropoff_long, passenger_count, dist_km]
|
63 |
+
# out_sz = 1
|
64 |
+
# layers = [200, 100]
|
65 |
+
# p = 0.4
|
66 |
+
|
67 |
+
# model = TabularModel(emb_szs, n_cont, out_sz, layers, p)
|
68 |
+
# # Load model state (using weights_only=True to address the warning)
|
69 |
+
# model.load_state_dict(torch.load("TaxiFareRegrModel.pt", map_location=torch.device("cpu"), weights_only=True))
|
70 |
+
# model.eval()
|
71 |
+
|
72 |
+
# # -----------------------------
|
73 |
+
# # Helper Function: Haversine
|
74 |
+
# # -----------------------------
|
75 |
+
# def haversine_distance_coords(lat1, lon1, lat2, lon2):
|
76 |
+
# """Compute haversine distance (in km) between two coordinate pairs."""
|
77 |
+
# r = 6371 # Earth's radius in kilometers
|
78 |
+
# phi1 = np.radians(lat1)
|
79 |
+
# phi2 = np.radians(lat2)
|
80 |
+
# delta_phi = np.radians(lat2 - lat1)
|
81 |
+
# delta_lambda = np.radians(lon2 - lon1)
|
82 |
+
# a = np.sin(delta_phi/2)**2 + np.cos(phi1)*np.cos(phi2)*np.sin(delta_lambda/2)**2
|
83 |
+
# c = 2 * np.arctan2(np.sqrt(a), np.sqrt(1 - a))
|
84 |
+
# return r * c
|
85 |
+
|
86 |
+
# # -----------------------------
|
87 |
+
# # OSRM Route Retrieval
|
88 |
+
# # -----------------------------
|
89 |
+
# def get_osrm_route(lat1, lon1, lat2, lon2):
|
90 |
+
# """
|
91 |
+
# Query OSRM for a route between (lat1, lon1) and (lat2, lon2).
|
92 |
+
# Returns:
|
93 |
+
# - route_points: list of (lat, lon) tuples for the route polyline
|
94 |
+
# - distance_m: route distance in meters (from OSRM)
|
95 |
+
# - duration_s: route duration in seconds (from OSRM)
|
96 |
+
# """
|
97 |
+
# # OSRM expects coordinates as "lon,lat;lon,lat"
|
98 |
+
# coords = f"{lon1},{lat1};{lon2},{lat2}"
|
99 |
+
# OSRM_URL = f"http://router.project-osrm.org/route/v1/driving/{coords}?overview=full&geometries=polyline"
|
100 |
+
|
101 |
+
# response = requests.get(OSRM_URL)
|
102 |
+
# response.raise_for_status()
|
103 |
+
# data = response.json()
|
104 |
+
|
105 |
+
# if data.get("code") != "Ok":
|
106 |
+
# raise Exception("Route not found")
|
107 |
+
|
108 |
+
# route = data["routes"][0]
|
109 |
+
# encoded_poly = route["geometry"]
|
110 |
+
# route_points = polyline.decode(encoded_poly)
|
111 |
+
# distance_m = route["distance"]
|
112 |
+
# duration_s = route["duration"]
|
113 |
+
# return route_points, distance_m, duration_s
|
114 |
+
|
115 |
+
# # -----------------------------
|
116 |
+
# # Main Prediction & Mapping Function
|
117 |
+
# # -----------------------------
|
118 |
+
# def predict_fare_and_map(plat, plong, dlat, dlong, psngr, dt):
|
119 |
+
# """
|
120 |
+
# 1. Process pickup datetime to extract categorical features.
|
121 |
+
# 2. Compute haversine distance for the model input.
|
122 |
+
# 3. Use the PyTorch model to predict the taxi fare.
|
123 |
+
# 4. Query OSRM for the actual road route geometry & distance.
|
124 |
+
# 5. Draw a Folium map with the OSRM route (blue line) and markers.
|
125 |
+
# 6. Return a text string with predicted fare and route distance, plus the map HTML.
|
126 |
+
# """
|
127 |
+
# # Process datetime
|
128 |
+
# try:
|
129 |
+
# pickup_dt = pd.to_datetime(dt)
|
130 |
+
# except Exception as e:
|
131 |
+
# return f"Error parsing date/time: {e}", ""
|
132 |
+
|
133 |
+
# hour = pickup_dt.hour
|
134 |
+
# am_or_pm = 0 if hour < 12 else 1
|
135 |
+
# weekday_str = pickup_dt.strftime("%a")
|
136 |
+
# weekday_map = {'Fri': 0, 'Mon': 1, 'Sat': 2, 'Sun': 3, 'Thu': 4, 'Tue': 5, 'Wed': 6}
|
137 |
+
# weekday = weekday_map.get(weekday_str, 0)
|
138 |
+
|
139 |
+
# # Prepare tensors for model input (use haversine distance)
|
140 |
+
# dist_km = haversine_distance_coords(plat, plong, dlat, dlong)
|
141 |
+
# cat_array = np.array([[hour, am_or_pm, weekday]])
|
142 |
+
# cat_tensor = torch.tensor(cat_array, dtype=torch.int64)
|
143 |
+
# cont_array = np.array([[plat, plong, dlat, dlong, psngr, dist_km]])
|
144 |
+
# cont_tensor = torch.tensor(cont_array, dtype=torch.float)
|
145 |
+
|
146 |
+
# # Predict fare
|
147 |
+
# with torch.no_grad():
|
148 |
+
# pred = model(cat_tensor, cont_tensor)
|
149 |
+
# fare_pred = pred.item()
|
150 |
+
|
151 |
+
# # Get route from OSRM
|
152 |
+
# try:
|
153 |
+
# route_points, route_distance_m, route_duration_s = get_osrm_route(plat, plong, dlat, dlong)
|
154 |
+
# except Exception as e:
|
155 |
+
# return f"Error from OSRM: {e}", ""
|
156 |
+
|
157 |
+
# # Create Folium map centered between pickup & dropoff
|
158 |
+
# mid_lat = (plat + dlat) / 2
|
159 |
+
# mid_lon = (plong + dlong) / 2
|
160 |
+
# m = folium.Map(location=[mid_lat, mid_lon], zoom_start=12)
|
161 |
+
|
162 |
+
# # Add markers
|
163 |
+
# folium.Marker([plat, plong], icon=folium.Icon(color="green"), tooltip="Pickup").add_to(m)
|
164 |
+
# folium.Marker([dlat, dlong], icon=folium.Icon(color="red"), tooltip="Dropoff").add_to(m)
|
165 |
+
|
166 |
+
# # Draw the route polyline (blue line) with popup showing OSRM distance
|
167 |
+
# folium.PolyLine(
|
168 |
+
# route_points,
|
169 |
+
# color="blue",
|
170 |
+
# weight=3,
|
171 |
+
# opacity=0.7,
|
172 |
+
# popup=f"OSRM Distance: {route_distance_m/1000:.2f} km"
|
173 |
+
# ).add_to(m)
|
174 |
+
|
175 |
+
# map_html = m._repr_html_()
|
176 |
+
|
177 |
+
# route_distance_km = route_distance_m / 1000
|
178 |
+
# output_text = (f"Predicted Fare: ${fare_pred:.2f}\n"
|
179 |
+
# f"Route Distance (OSRM): {route_distance_km:.2f} km")
|
180 |
+
# return output_text, map_html
|
181 |
+
|
182 |
+
# # -----------------------------
|
183 |
+
# # Gradio Interface
|
184 |
+
# # -----------------------------
|
185 |
+
# iface = gr.Interface(
|
186 |
+
# fn=predict_fare_and_map,
|
187 |
+
# inputs=[
|
188 |
+
# gr.Number(label="Pickup Latitude", value=40.75),
|
189 |
+
# gr.Number(label="Pickup Longitude", value=-73.99),
|
190 |
+
# gr.Number(label="Dropoff Latitude", value=40.73),
|
191 |
+
# gr.Number(label="Dropoff Longitude", value=-73.98),
|
192 |
+
# gr.Number(label="Passenger Count", value=1),
|
193 |
+
# gr.Textbox(label="Pickup Date and Time (YYYY-MM-DD HH:MM:SS)", value="2010-04-19 08:17:56")
|
194 |
+
# ],
|
195 |
+
# outputs=[
|
196 |
+
# gr.Textbox(label="Prediction & Distance"),
|
197 |
+
# gr.HTML(label="Map")
|
198 |
+
# ],
|
199 |
+
# title="NYC Taxi Fare Prediction with OSRM Road Route",
|
200 |
+
# description=(
|
201 |
+
# "Enter pickup/dropoff coordinates, passenger count, and pickup datetime to predict the taxi fare. "
|
202 |
+
# "The app displays the actual road route (blue line) from OSRM on a Folium map."
|
203 |
+
# )
|
204 |
+
# )
|
205 |
+
|
206 |
+
# if __name__ == "__main__":
|
207 |
+
# iface.launch()
|
208 |
+
|
209 |
#!/usr/bin/env python
|
210 |
"""
|
211 |
Gradio App for NYC Taxi Fare Prediction & Road Route Visualization using OSRM
|
|
|
387 |
f"Route Distance (OSRM): {route_distance_km:.2f} km")
|
388 |
return output_text, map_html
|
389 |
|
390 |
+
# -----------------------------
|
391 |
+
# Example Locations (Popular NYC Spots)
|
392 |
+
# Each example is a list of 6 inputs:
|
393 |
+
# [pickup_lat, pickup_lon, dropoff_lat, dropoff_lon, passenger_count, pickup_datetime]
|
394 |
+
# -----------------------------
|
395 |
+
examples = [
|
396 |
+
# 1. Times Square to Central Park (short ride)
|
397 |
+
[40.7580, -73.9855, 40.7690, -73.9819, 1, "2010-04-19 08:17:56"],
|
398 |
+
# 2. Times Square to JFK Airport (long ride)
|
399 |
+
[40.7580, -73.9855, 40.6413, -73.7781, 1, "2010-04-19 08:17:56"],
|
400 |
+
# 3. Grand Central Terminal to Empire State Building (very short ride)
|
401 |
+
[40.7527, -73.9772, 40.7484, -73.9857, 1, "2010-04-19 08:17:56"],
|
402 |
+
# 4. Brooklyn Bridge to Wall Street (short urban ride)
|
403 |
+
[40.7061, -73.9969, 40.7069, -74.0113, 1, "2010-04-19 08:17:56"],
|
404 |
+
# 5. Yankee Stadium to Central Park (moderate ride)
|
405 |
+
[40.8296, -73.9262, 40.7829, -73.9654, 1, "2010-04-19 08:17:56"],
|
406 |
+
# 6. Columbia University area to Rockefeller Center (cross-city ride)
|
407 |
+
[40.8075, -73.9626, 40.7587, -73.9787, 1, "2010-04-19 08:17:56"],
|
408 |
+
# 7. Battery Park to Central Park (longer ride across Manhattan)
|
409 |
+
[40.7033, -74.0170, 40.7829, -73.9654, 1, "2010-04-19 08:17:56"]
|
410 |
+
]
|
411 |
+
|
412 |
# -----------------------------
|
413 |
# Gradio Interface
|
414 |
# -----------------------------
|
|
|
426 |
gr.Textbox(label="Prediction & Distance"),
|
427 |
gr.HTML(label="Map")
|
428 |
],
|
429 |
+
examples=examples,
|
430 |
title="NYC Taxi Fare Prediction with OSRM Road Route",
|
431 |
description=(
|
432 |
"Enter pickup/dropoff coordinates, passenger count, and pickup datetime to predict the taxi fare. "
|
433 |
+
"The app displays the actual road route (blue line) from OSRM on a Folium map. "
|
434 |
+
"You can also choose from several example routes between popular locations in New York."
|
435 |
)
|
436 |
)
|
437 |
|
438 |
if __name__ == "__main__":
|
439 |
iface.launch()
|
440 |
+
|