Spaces:
Sleeping
Sleeping
Upload 5 files
Browse files- img/taxi_img.png +0 -0
- main.py +203 -0
- models/min_max_scaler.pkl +3 -0
- models/model_xgb.pkl +3 -0
- requirements.txt +9 -0
img/taxi_img.png
ADDED
main.py
ADDED
@@ -0,0 +1,203 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import requests
|
3 |
+
import plotly.graph_objects as go
|
4 |
+
from geopy.geocoders import Nominatim
|
5 |
+
import pandas as pd
|
6 |
+
from datetime import datetime
|
7 |
+
import holidays
|
8 |
+
import numpy as np
|
9 |
+
from sklearn.preprocessing import MinMaxScaler
|
10 |
+
import pickle
|
11 |
+
import xgboost as xgb
|
12 |
+
|
13 |
+
# Setting up the page configuration for Streamlit App
|
14 |
+
st.set_page_config(
|
15 |
+
page_title="Taxi",
|
16 |
+
# layout="wide",
|
17 |
+
initial_sidebar_state="expanded"
|
18 |
+
)
|
19 |
+
|
20 |
+
# Load the XGBoost model
|
21 |
+
#@st.cache_data()
|
22 |
+
def get_model():
|
23 |
+
model = pickle.load(open("models/model_xgb.pkl", "rb"))
|
24 |
+
return model
|
25 |
+
|
26 |
+
# Function to make prediction using the model and input data
|
27 |
+
def make_prediction(data):
|
28 |
+
model = get_model()
|
29 |
+
best_features = ['vendor_id', 'passenger_count', 'pickup_longitude', 'pickup_latitude',
|
30 |
+
'dropoff_longitude', 'dropoff_latitude', 'store_and_fwd_flag',
|
31 |
+
'pickup_hour', 'pickup_holiday', 'total_distance', 'total_travel_time',
|
32 |
+
'number_of_steps', 'haversine_distance', 'temperature',
|
33 |
+
'pickup_day_of_week_1', 'pickup_day_of_week_2', 'pickup_day_of_week_3',
|
34 |
+
'pickup_day_of_week_4', 'pickup_day_of_week_5', 'pickup_day_of_week_6',
|
35 |
+
'geo_cluster_1', 'geo_cluster_3', 'geo_cluster_5', 'geo_cluster_7',
|
36 |
+
'geo_cluster_9']
|
37 |
+
data_matrix = xgb.DMatrix(data, feature_names=best_features)
|
38 |
+
return model.predict(data_matrix)
|
39 |
+
|
40 |
+
|
41 |
+
def get_coordinates(address):
|
42 |
+
# Создание экземпляра геокодера
|
43 |
+
geolocator = Nominatim(user_agent="my_app")
|
44 |
+
|
45 |
+
# Получение координат по адресу
|
46 |
+
location = geolocator.geocode(address)
|
47 |
+
|
48 |
+
# Вывод широты и долготы
|
49 |
+
return (location.longitude, location.latitude)
|
50 |
+
|
51 |
+
|
52 |
+
def show_map(lon_from, lat_from, lon_to, lat_to):
|
53 |
+
# Создание карты
|
54 |
+
fig = go.Figure(go.Scattermapbox(
|
55 |
+
mode = "markers",
|
56 |
+
marker = {'size': 15, 'color': 'red'}
|
57 |
+
))
|
58 |
+
|
59 |
+
# Добавление флажков для точек
|
60 |
+
fig.add_trace(go.Scattermapbox(
|
61 |
+
mode = "markers",
|
62 |
+
lon = [lon_from, lon_to],
|
63 |
+
lat = [lat_from, lat_to],
|
64 |
+
marker = go.scattermapbox.Marker(
|
65 |
+
size=25,
|
66 |
+
color='red'
|
67 |
+
)
|
68 |
+
))
|
69 |
+
|
70 |
+
# Добавление линии между точками
|
71 |
+
fig.add_trace(go.Scattermapbox(
|
72 |
+
mode = "lines",
|
73 |
+
lon = [lon_from, lon_to],
|
74 |
+
lat = [lat_from, lat_to],
|
75 |
+
line = dict(width=2, color='green')
|
76 |
+
))
|
77 |
+
|
78 |
+
# Настройка отображения карты
|
79 |
+
fig.update_layout(
|
80 |
+
mapbox = {
|
81 |
+
'style': "open-street-map", # Стиль карты
|
82 |
+
'center': {'lon': (lon_from + lon_to) / 2, 'lat': (lat_from + lat_to) / 2}, # Центр карты
|
83 |
+
'zoom': 9, # Уровень масштабирования карты
|
84 |
+
},
|
85 |
+
showlegend = False,
|
86 |
+
height = 600, # Изменение высоты карты
|
87 |
+
width = 1200 # Изменение ширины карты
|
88 |
+
)
|
89 |
+
|
90 |
+
# Отображение карты
|
91 |
+
return fig
|
92 |
+
|
93 |
+
|
94 |
+
# Get total distance
|
95 |
+
def get_total_distance(start_longitude, start_latitude, end_longitude, end_latitude):
|
96 |
+
# Construct the URL for sending a request to the public OSRM server
|
97 |
+
url = f"http://router.project-osrm.org/route/v1/driving/{start_longitude},{start_latitude};{end_longitude},{end_latitude}?overview=false"
|
98 |
+
|
99 |
+
# Send a GET request to the OSRM server
|
100 |
+
response = requests.get(url)
|
101 |
+
|
102 |
+
# Process the response from the server
|
103 |
+
if response.status_code == 200:
|
104 |
+
data = response.json()
|
105 |
+
total_distance = data["routes"][0]["distance"] # Total distance in meters
|
106 |
+
total_travel_time = data["routes"][0]["duration"] # Total travel time in seconds
|
107 |
+
number_of_steps = len(data["routes"][0]["legs"][0]["steps"]) # Number of steps in the
|
108 |
+
return total_distance, total_travel_time, number_of_steps
|
109 |
+
|
110 |
+
|
111 |
+
# Get Harversine distance
|
112 |
+
def get_haversine_distance(lat1, lng1, lat2, lng2):
|
113 |
+
# Convert angles to radians
|
114 |
+
lat1, lng1, lat2, lng2 = map(np.radians, (lat1, lng1, lat2, lng2))
|
115 |
+
# Earth's radius in kilometers
|
116 |
+
EARTH_RADIUS = 6371
|
117 |
+
# Calculate the shortest distance h using the Haversine formula
|
118 |
+
lat_delta = lat2 - lat1
|
119 |
+
lng_delta = lng2 - lng1
|
120 |
+
d = np.sin(lat_delta * 0.5) ** 2 + np.cos(lat1) * np.cos(lat2) * np.sin(lng_delta * 0.5) ** 2
|
121 |
+
h = 2 * EARTH_RADIUS * np.arcsin(np.sqrt(d))
|
122 |
+
return h
|
123 |
+
|
124 |
+
|
125 |
+
# User input features
|
126 |
+
def user_input_features(lon_from, lat_from, lon_to, lat_to, passenger_count, temperature):
|
127 |
+
current_time = datetime.now()
|
128 |
+
pickup_hour= current_time.hour
|
129 |
+
today = datetime.today()
|
130 |
+
pickup_holiday = 1 if today in holidays.USA() else 0
|
131 |
+
total_distance, total_travel_time, number_of_steps = get_total_distance(lon_from, lat_from, lon_to, lat_to)
|
132 |
+
haversine_distance = get_haversine_distance(lat_from, lon_from, lat_to, lon_to)
|
133 |
+
weekday_number = current_time.weekday()
|
134 |
+
|
135 |
+
data = {'vendor_id': 1,
|
136 |
+
'passenger_count': passenger_count,
|
137 |
+
'pickup_longitude': lon_from,
|
138 |
+
'pickup_latitude': lat_from,
|
139 |
+
'dropoff_longitude': lon_to,
|
140 |
+
'dropoff_latitude': lat_to,
|
141 |
+
'store_and_fwd_flag': 0.0,
|
142 |
+
'pickup_hour': pickup_hour,
|
143 |
+
'pickup_holiday': pickup_holiday,
|
144 |
+
'total_distance': total_distance,
|
145 |
+
'total_travel_time': total_travel_time,
|
146 |
+
'number_of_steps': number_of_steps,
|
147 |
+
'haversine_distance': haversine_distance,
|
148 |
+
'temperature': temperature,
|
149 |
+
'pickup_day_of_week_1': 1 if weekday_number == 1 else 0,
|
150 |
+
'pickup_day_of_week_2': 1 if weekday_number == 2 else 0,
|
151 |
+
'pickup_day_of_week_3': 1 if weekday_number == 3 else 0,
|
152 |
+
'pickup_day_of_week_4': 1 if weekday_number == 4 else 0,
|
153 |
+
'pickup_day_of_week_5': 1 if weekday_number == 5 else 0,
|
154 |
+
'pickup_day_of_week_6': 1 if weekday_number == 6 else 0,
|
155 |
+
'geo_cluster_1':1,
|
156 |
+
'geo_cluster_3':0,
|
157 |
+
'geo_cluster_5':0,
|
158 |
+
'geo_cluster_7':0,
|
159 |
+
'geo_cluster_9':0
|
160 |
+
}
|
161 |
+
features = pd.DataFrame(data, index=[0])
|
162 |
+
return features
|
163 |
+
|
164 |
+
|
165 |
+
# Scale the input data using a pre-trained MinMaxScaler
|
166 |
+
def min_max_scaler(data):
|
167 |
+
scaler = pickle.load(open("models/min_max_scaler.pkl", "rb"))
|
168 |
+
data_scaled = scaler.transform(data)
|
169 |
+
return data_scaled
|
170 |
+
|
171 |
+
# Main function
|
172 |
+
def main():
|
173 |
+
|
174 |
+
if 'btn_predict' not in st.session_state:
|
175 |
+
st.session_state['btn_predict'] = False
|
176 |
+
|
177 |
+
# Sidebar
|
178 |
+
st.sidebar.markdown(''' # New York City Taxi Trip Duration''')
|
179 |
+
st.sidebar.image("img/taxi_img.png")
|
180 |
+
address_from = st.sidebar.text_input("Откуда:", value="New York, Liberty Island")
|
181 |
+
address_to = st.sidebar.text_input("Куда:", value="New York, 20 W 34th St")
|
182 |
+
passenger_count = st.sidebar.slider("Количество пассажиров", 1, 4, 1)
|
183 |
+
temperature = st.sidebar.slider("Temperature (C)", -20, 40, 15)
|
184 |
+
st.session_state['btn_predict'] = st.sidebar.button('Start')
|
185 |
+
|
186 |
+
if st.session_state['btn_predict']:
|
187 |
+
lon_from, lat_from = get_coordinates(address_from)
|
188 |
+
lon_to, lat_to = get_coordinates(address_to)
|
189 |
+
st.plotly_chart(show_map(lon_from, lat_from, lon_to, lat_to))
|
190 |
+
user_data = user_input_features(lon_from, lat_from, lon_to, lat_to, passenger_count, temperature)
|
191 |
+
# st.write(user_data)
|
192 |
+
data_scaled = min_max_scaler(user_data)
|
193 |
+
trip_duration = np.exp(make_prediction(data_scaled)) - 1
|
194 |
+
trip_duration = round(float(trip_duration) / 60)
|
195 |
+
st.markdown(f"""
|
196 |
+
<div style='background-color: lightgreen; padding: 10px;'>
|
197 |
+
<h2 style='color: black; text-align: center;'>Длительность поездки составит: {trip_duration} мин.</h2>
|
198 |
+
</div>
|
199 |
+
""", unsafe_allow_html=True)
|
200 |
+
|
201 |
+
# Running the main function
|
202 |
+
if __name__ == "__main__":
|
203 |
+
main()
|
models/min_max_scaler.pkl
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2974581d9e870affcb1eaa0eb86290630840e6262eddd70d604f8415a789493a
|
3 |
+
size 2036
|
models/model_xgb.pkl
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:83d6e22ee287b9a2e15efefbebb5376b89815f6b06212eb48a338ad745a046bf
|
3 |
+
size 1330844
|
requirements.txt
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
streamlit
|
2 |
+
requests
|
3 |
+
plotly
|
4 |
+
geopy
|
5 |
+
datetime
|
6 |
+
pandas
|
7 |
+
holidays
|
8 |
+
scikit-learn
|
9 |
+
xgboost
|