Upload 4 files

app.py

@@ -0,0 +1,97 @@
+import gradio as gr
+import pandas as pd
+import joblib
+import numpy as np
+
+# Modeli yükle
+model = joblib.load("final_lgb_nyc.pkl")  # senin final modeli
+
+# Örnek lokasyon listesi (mahalle/sokak)
+locations = [
+    "Manhattan", "Brooklyn", "Queens", "Bronx", "Staten Island"
+]
+
+# Örnek hava durumu seçenekleri
+weather_options = ["Clear", "Rain", "Snow", "Fog", "Cloudy"]
+
+def predict_trip_duration(
+    pickup_location, dropoff_location, pickup_hour, pickup_dayofweek,
+    pickup_month, weather, passenger_count
+):
+    # Basit feature engineering
+    vendor_id = 1  # sabit varsayım
+    store_and_fwd_flag = 0  # sabit varsayım
+    
+    # Örnek koordinat ataması (gerçek uygulamada bir lookup tablosu gerekir)
+    loc_coords = {
+        "Manhattan": (40.7580, -73.9855),
+        "Brooklyn": (40.6782, -73.9442),
+        "Queens": (40.7282, -73.7949),
+        "Bronx": (40.8448, -73.8648),
+        "Staten Island": (40.5795, -74.1502)
+    }
+    
+    pickup_lat, pickup_lon = loc_coords[pickup_location]
+    dropoff_lat, dropoff_lon = loc_coords[dropoff_location]
+    
+    # Mesafe feature’leri
+    manhattan_distance = abs(pickup_lon - dropoff_lon) + abs(pickup_lat - dropoff_lat)
+    euclidean_distance = np.sqrt((pickup_lon - dropoff_lon)**2 + (pickup_lat - dropoff_lat)**2)
+    longitude_diff = abs(pickup_lon - dropoff_lon)
+    latitude_diff = abs(pickup_lat - dropoff_lat)
+    
+    speed_estimate = 0  # placeholder
+    passenger_per_distance = passenger_count / (manhattan_distance + 0.001)
+    
+    # Categorical encode
+    is_weekend = int(pickup_dayofweek in [5,6])
+    
+    weather_mapping = {w:i for i,w in enumerate(weather_options)}
+    weather_encoded = weather_mapping.get(weather, 0)
+    
+    df = pd.DataFrame({
+        "pickup_hour": [pickup_hour],
+        "pickup_dayofweek": [pickup_dayofweek],
+        "pickup_month": [pickup_month],
+        "manhattan_distance": [manhattan_distance],
+        "euclidean_distance": [euclidean_distance],
+        "longitude_diff": [longitude_diff],
+        "latitude_diff": [latitude_diff],
+        "speed_estimate": [speed_estimate],
+        "passenger_per_distance": [passenger_per_distance],
+        "vendor_id": [vendor_id],
+        "store_and_fwd_flag": [store_and_fwd_flag],
+        "is_weekend": [is_weekend],
+        "weather": [weather_encoded]
+    })
+    
+    # Tahmin (log-transform modeli ise expm1 uygulanabilir)
+    pred_log = model.predict(df)
+    pred = np.expm1(pred_log) if np.min(pred_log) > 0 else pred_log
+    pred = max(pred[0], 0)  # negatif tahminleri 0 yap
+    
+    return f"{pred:.2f} saniye (~{pred/60:.1f} dakika)"
+
+# --- Arayüz inputları ---
+inputs = [
+    gr.Dropdown(label="Pickup Location", choices=locations, value="Manhattan"),
+    gr.Dropdown(label="Dropoff Location", choices=locations, value="Brooklyn"),
+    gr.Slider(label="Pickup Hour", minimum=0, maximum=23, step=1, value=9),
+    gr.Slider(label="Pickup Day of Week (0=Mon)", minimum=0, maximum=6, step=1, value=1),
+    gr.Slider(label="Pickup Month", minimum=1, maximum=12, step=1, value=6),
+    gr.Dropdown(label="Weather", choices=weather_options, value="Clear"),
+    gr.Slider(label="Passenger Count", minimum=1, maximum=6, step=1, value=1)
+]
+
+# --- Output ---
+outputs = gr.Textbox(label="Tahmini Trip Duration")
+
+iface = gr.Interface(
+    fn=predict_trip_duration,
+    inputs=inputs,
+    outputs=outputs,
+    title="NYC Taxi Trip Duration Predictor",
+    description="Kullanıcının girdiği lokasyon, saat ve hava durumuna göre tahmini yolculuk süresini verir."
+)
+
+iface.launch()

nyc_lgb_best_model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c405620739062cc3162a18b12c88a1fbcf2058db4c630e065b3b3b397d655d6
+size 7171742

requirements.txt

@@ -0,0 +1,9 @@
+pandas
+numpy
+lightgbm
+xgboost
+joblib
+shap
+scikit-learn
+matplotlib
+gradio