Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -4,32 +4,30 @@ API de prédiction du prix de location - Projet Getaround
|
|
| 4 |
Cette API expose un point de terminaison /predict qui renvoie le prix
|
| 5 |
de location journalier suggéré pour un véhicule, à partir de ses
|
| 6 |
caractéristiques.
|
| 7 |
-
|
| 8 |
-
Auteur : Projet déploiement
|
| 9 |
"""
|
| 10 |
|
| 11 |
from typing import List, Union
|
| 12 |
from pathlib import Path
|
| 13 |
import joblib
|
| 14 |
-
import numpy as np
|
| 15 |
import pandas as pd
|
| 16 |
from fastapi import FastAPI, HTTPException
|
| 17 |
-
from fastapi.responses import HTMLResponse
|
| 18 |
from pydantic import BaseModel, Field
|
| 19 |
|
| 20 |
# ---------------------------------------------------------------------------
|
| 21 |
# Configuration de l'application
|
| 22 |
# ---------------------------------------------------------------------------
|
| 23 |
-
APP_TITLE = "Getaround -
|
| 24 |
-
APP_DESCRIPTION =
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
)
|
|
|
|
|
|
|
|
|
|
| 29 |
APP_VERSION = "1.0.0"
|
| 30 |
MODEL_PATH = Path(__file__).resolve().parent / "best_model.joblib"
|
| 31 |
|
| 32 |
-
# Colonnes attendues dans l'ordre du modèle
|
| 33 |
FEATURE_COLUMNS = [
|
| 34 |
"model_key", "mileage", "engine_power", "fuel", "paint_color",
|
| 35 |
"car_type", "private_parking_available", "has_gps",
|
|
@@ -37,29 +35,16 @@ FEATURE_COLUMNS = [
|
|
| 37 |
"has_speed_regulator", "winter_tires"
|
| 38 |
]
|
| 39 |
|
| 40 |
-
# Modalités acceptées (à titre informatif)
|
| 41 |
-
ACCEPTED_MODELS = [
|
| 42 |
-
"Citroën", "Renault", "BMW", "Peugeot", "Audi", "Nissan", "Mitsubishi",
|
| 43 |
-
"Mercedes", "Volkswagen", "Toyota", "SEAT", "Subaru", "Opel", "PGO",
|
| 44 |
-
"Ferrari", "Other"
|
| 45 |
-
]
|
| 46 |
-
ACCEPTED_FUELS = ["diesel", "petrol", "hybrid_petrol", "electro"]
|
| 47 |
-
ACCEPTED_COLORS = ["black", "grey", "white", "red", "silver", "blue",
|
| 48 |
-
"orange", "beige", "brown", "green"]
|
| 49 |
-
ACCEPTED_CAR_TYPES = ["convertible", "coupe", "estate", "hatchback",
|
| 50 |
-
"sedan", "subcompact", "suv", "van"]
|
| 51 |
-
|
| 52 |
# ---------------------------------------------------------------------------
|
| 53 |
# Initialisation
|
| 54 |
# ---------------------------------------------------------------------------
|
| 55 |
app = FastAPI(
|
| 56 |
title=APP_TITLE,
|
| 57 |
description=APP_DESCRIPTION,
|
| 58 |
-
version=APP_VERSION
|
| 59 |
-
redoc_url="/redoc" # On laisse FastAPI gérer automatiquement Swagger sur /docs
|
| 60 |
)
|
| 61 |
|
| 62 |
-
# Chargement du modèle
|
| 63 |
try:
|
| 64 |
model = joblib.load(MODEL_PATH)
|
| 65 |
MODEL_LOADED = True
|
|
@@ -72,7 +57,6 @@ except Exception as exc:
|
|
| 72 |
# Schémas Pydantic
|
| 73 |
# ---------------------------------------------------------------------------
|
| 74 |
class PredictionInput(BaseModel):
|
| 75 |
-
"""Schéma d'entrée acceptant une liste de listes de caractéristiques."""
|
| 76 |
input: List[List[Union[str, int, float, bool]]] = Field(
|
| 77 |
...,
|
| 78 |
examples=[[[
|
|
@@ -89,13 +73,12 @@ class PredictionOutput(BaseModel):
|
|
| 89 |
# ---------------------------------------------------------------------------
|
| 90 |
@app.get("/", tags=["Accueil"])
|
| 91 |
async def root():
|
| 92 |
-
"""Point d'entrée par défaut.
|
| 93 |
return {
|
| 94 |
"message": "API Getaround opérationnelle",
|
| 95 |
"version": APP_VERSION,
|
| 96 |
"model_loaded": MODEL_LOADED,
|
| 97 |
-
"documentation": "/docs"
|
| 98 |
-
"presentation_html": "/about"
|
| 99 |
}
|
| 100 |
|
| 101 |
@app.get("/health", tags=["Accueil"])
|
|
@@ -110,31 +93,24 @@ async def health():
|
|
| 110 |
async def predict(payload: PredictionInput):
|
| 111 |
"""Renvoie la prédiction du prix de location journalier (EUR/jour)."""
|
| 112 |
if not MODEL_LOADED:
|
| 113 |
-
raise HTTPException(status_code=503,
|
| 114 |
-
detail="Modèle non chargé sur le serveur.")
|
| 115 |
try:
|
| 116 |
df_input = pd.DataFrame(payload.input, columns=FEATURE_COLUMNS)
|
| 117 |
|
| 118 |
-
# Conversion
|
| 119 |
-
bool_cols = ["private_parking_available", "has_gps",
|
| 120 |
-
"
|
| 121 |
-
"has_getaround_connect", "has_speed_regulator",
|
| 122 |
-
"winter_tires"]
|
| 123 |
for c in bool_cols:
|
| 124 |
df_input[c] = df_input[c].astype(int)
|
| 125 |
|
| 126 |
-
# Conversion des numériques
|
| 127 |
df_input["mileage"] = df_input["mileage"].astype(int)
|
| 128 |
df_input["engine_power"] = df_input["engine_power"].astype(int)
|
| 129 |
|
| 130 |
preds = model.predict(df_input)
|
| 131 |
return {"prediction": [round(float(p), 2) for p in preds]}
|
| 132 |
except Exception as exc:
|
| 133 |
-
raise HTTPException(status_code=400,
|
| 134 |
-
detail=f"Erreur lors de la prédiction : {exc}")
|
| 135 |
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
@app.get("/about", response_class=HTMLResponse, include_in_schema=False)
|
| 140 |
-
async def custom_
|
|
|
|
| 4 |
Cette API expose un point de terminaison /predict qui renvoie le prix
|
| 5 |
de location journalier suggéré pour un véhicule, à partir de ses
|
| 6 |
caractéristiques.
|
|
|
|
|
|
|
| 7 |
"""
|
| 8 |
|
| 9 |
from typing import List, Union
|
| 10 |
from pathlib import Path
|
| 11 |
import joblib
|
|
|
|
| 12 |
import pandas as pd
|
| 13 |
from fastapi import FastAPI, HTTPException
|
|
|
|
| 14 |
from pydantic import BaseModel, Field
|
| 15 |
|
| 16 |
# ---------------------------------------------------------------------------
|
| 17 |
# Configuration de l'application
|
| 18 |
# ---------------------------------------------------------------------------
|
| 19 |
+
APP_TITLE = "API Getaround - Prédiction du prix de location"
|
| 20 |
+
APP_DESCRIPTION = """
|
| 21 |
+
Cette API met à disposition un modèle de Machine Learning entraîné pour suggérer le prix journalier optimal d'une location de véhicule sur la plateforme Getaround.
|
| 22 |
+
|
| 23 |
+
### Performances du modèle (XGBoost Regressor) :
|
| 24 |
+
* **Erreur absolue moyenne (MAE) :** 9,18 EUR
|
| 25 |
+
* **RMSE :** 12,82 EUR
|
| 26 |
+
* **R² :** 0,846
|
| 27 |
+
"""
|
| 28 |
APP_VERSION = "1.0.0"
|
| 29 |
MODEL_PATH = Path(__file__).resolve().parent / "best_model.joblib"
|
| 30 |
|
|
|
|
| 31 |
FEATURE_COLUMNS = [
|
| 32 |
"model_key", "mileage", "engine_power", "fuel", "paint_color",
|
| 33 |
"car_type", "private_parking_available", "has_gps",
|
|
|
|
| 35 |
"has_speed_regulator", "winter_tires"
|
| 36 |
]
|
| 37 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
# ---------------------------------------------------------------------------
|
| 39 |
# Initialisation
|
| 40 |
# ---------------------------------------------------------------------------
|
| 41 |
app = FastAPI(
|
| 42 |
title=APP_TITLE,
|
| 43 |
description=APP_DESCRIPTION,
|
| 44 |
+
version=APP_VERSION
|
|
|
|
| 45 |
)
|
| 46 |
|
| 47 |
+
# Chargement du modèle
|
| 48 |
try:
|
| 49 |
model = joblib.load(MODEL_PATH)
|
| 50 |
MODEL_LOADED = True
|
|
|
|
| 57 |
# Schémas Pydantic
|
| 58 |
# ---------------------------------------------------------------------------
|
| 59 |
class PredictionInput(BaseModel):
|
|
|
|
| 60 |
input: List[List[Union[str, int, float, bool]]] = Field(
|
| 61 |
...,
|
| 62 |
examples=[[[
|
|
|
|
| 73 |
# ---------------------------------------------------------------------------
|
| 74 |
@app.get("/", tags=["Accueil"])
|
| 75 |
async def root():
|
| 76 |
+
"""Point d'entrée par défaut."""
|
| 77 |
return {
|
| 78 |
"message": "API Getaround opérationnelle",
|
| 79 |
"version": APP_VERSION,
|
| 80 |
"model_loaded": MODEL_LOADED,
|
| 81 |
+
"documentation": "/docs"
|
|
|
|
| 82 |
}
|
| 83 |
|
| 84 |
@app.get("/health", tags=["Accueil"])
|
|
|
|
| 93 |
async def predict(payload: PredictionInput):
|
| 94 |
"""Renvoie la prédiction du prix de location journalier (EUR/jour)."""
|
| 95 |
if not MODEL_LOADED:
|
| 96 |
+
raise HTTPException(status_code=503, detail="Modèle non chargé sur le serveur.")
|
|
|
|
| 97 |
try:
|
| 98 |
df_input = pd.DataFrame(payload.input, columns=FEATURE_COLUMNS)
|
| 99 |
|
| 100 |
+
# Conversion des types
|
| 101 |
+
bool_cols = ["private_parking_available", "has_gps", "has_air_conditioning",
|
| 102 |
+
"automatic_car", "has_getaround_connect", "has_speed_regulator", "winter_tires"]
|
|
|
|
|
|
|
| 103 |
for c in bool_cols:
|
| 104 |
df_input[c] = df_input[c].astype(int)
|
| 105 |
|
|
|
|
| 106 |
df_input["mileage"] = df_input["mileage"].astype(int)
|
| 107 |
df_input["engine_power"] = df_input["engine_power"].astype(int)
|
| 108 |
|
| 109 |
preds = model.predict(df_input)
|
| 110 |
return {"prediction": [round(float(p), 2) for p in preds]}
|
| 111 |
except Exception as exc:
|
| 112 |
+
raise HTTPException(status_code=400, detail=f"Erreur lors de la prédiction : {exc}")
|
|
|
|
| 113 |
|
| 114 |
+
if __name__ == "__main__":
|
| 115 |
+
import uvicorn
|
| 116 |
+
uvicorn.run("app:app", host="0.0.0.0", port=8000, reload=False)
|
|
|
|
|
|