first

Dockerfile

@@ -0,0 +1,15 @@
+FROM python:3.11-slim
+
+WORKDIR /app
+
+# Copier tous les fichiers
+COPY . /app
+
+# Installer les dépendances
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Exposer le port attendu par HF
+EXPOSE 7860
+
+# Lancer FastAPI
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -0,0 +1,116 @@
+from fastapi import FastAPI, Body, Request
+from fastapi.responses import JSONResponse, RedirectResponse
+from fastapi.exceptions import RequestValidationError
+from pydantic import BaseModel, Field
+from typing import Literal
+import pandas as pd
+import joblib
+import uvicorn
+
+# Charger le modèle
+model = joblib.load("get_around_v1.pkl")
+
+# Classe Pydantic pour les entrées
+class CarData(BaseModel):
+    mileage: float = Field(..., ge=0, description="Kilométrage doit être >= 0")
+    engine_power: float = Field(..., ge=1, le=423, description="Puissance moteur entre 1 et 423")
+    car_brand: Literal[
+        "Citroën", "Peugeot", "PGO", "Renault", "Audi", "BMW", "Ford", "Mercedes",
+        "Opel", "Porsche", "Volkswagen", "KIA Motors", "Alfa Romeo", "Ferrari",
+        "Fiat", "Lamborghini", "Maserati", "Lexus", "Honda", "Mazda", "Mini",
+        "Mitsubishi", "Nissan", "SEAT", "Subaru", "Suzuki", "Toyota", "Yamaha"
+    ]
+    fuel: Literal["diesel", "petrol", "hybrid_petrol", "electro"]
+    paint_color: Literal[
+        "black", "grey", "white", "red", "silver", "blue",
+        "orange", "beige", "brown", "green"
+    ]
+    car_type: Literal[
+        "convertible", "coupe", "estate", "hatchback", "sedan", "subcompact", "suv", "van"
+    ]
+    private_parking_available: bool
+    has_gps: bool
+    has_air_conditioning: bool
+    automatic_car: bool
+    has_getaround_connect: bool
+    has_speed_regulator: bool
+    winter_tires: bool
+
+# Initialisation FastAPI
+app = FastAPI(
+    title="GetAround Prediction API",
+    description="API de prédiction du prix de location par jour pour GetAround",
+    version="1.0"
+)
+
+
+@app.get("/")
+def root():
+    return RedirectResponse(url="/docs")
+
+# Gestionnaire d'erreurs personnalisé
+@app.exception_handler(RequestValidationError)
+async def validation_exception_handler(request: Request, exc: RequestValidationError):
+    errors = []
+    for e in exc.errors():
+        field = e.get("loc")[-1]
+        expected = e.get("ctx", {}).get("expected")
+        msg = e.get("msg")
+        if expected:
+            errors.append({
+                "field": field,
+                "message": f"Valeur incorrecte pour '{field}'. Valeurs possibles : {expected}"
+            })
+        else:
+            errors.append({
+                "field": field,
+                "message": f"Erreur sur '{field}': {msg}"
+            })
+    return JSONResponse(
+        status_code=422,
+        content={
+            "error": "Certaines valeurs ne sont pas valides",
+            "details": errors
+        }
+    )
+
+# Endpoint de prédiction
+@app.post(
+    "/predict",
+    summary="Prédire le prix journalier d'une voiture",
+    description="Fournir toutes les caractéristiques de la voiture pour obtenir le prix prédictif."
+)
+def predict(
+    data: CarData = Body(
+        ...,
+        examples={
+            "valid_example": {
+                "summary": "Exemple valide",
+                "value": {
+                    "mileage": 15000,
+                    "engine_power": 100,
+                    "car_brand": "Citroën",
+                    "fuel": "diesel",
+                    "paint_color": "white",
+                    "car_type": "sedan",
+                    "private_parking_available": True,
+                    "has_gps": True,
+                    "has_air_conditioning": True,
+                    "automatic_car": True,
+                    "has_getaround_connect": True,
+                    "has_speed_regulator": True,
+                    "winter_tires": True
+                }
+            }
+        }
+    )
+):
+    df = pd.DataFrame([data.model_dump()])
+    df.rename(columns={'car_brand': 'model_key'}, inplace=True)
+    prediction = model.predict(df)
+    return {"prediction": round(float(prediction[0]), 2)}
+
+
+# Lancer localement (utile pour tests)
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=7860)

get_around_v1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84b6fe653ddf116a6b1708e6a5bebba58f2655ca0dfe49ac96b7775ae503556f
+size 7861686

requirements.txt

@@ -0,0 +1,12 @@
+fastapi==0.111.1
+pydantic==2.8.0
+uvicorn==0.23.2
+pandas==2.1.1
+scikit-learn==1.5.1
+joblib==1.3.2
+typing-extensions==4.9.0
+requests==2.32.0
+gradio==3.44.0
+xgboost
+gradio
+websockets