---
license: mit
language:
- en
---

# BMW Model Predictor

This model predicts the BMW car model using XGBoost based on price, mileage, fuel type, engine power, and car type.
---
license: mit
language:
  - en
tags:
  - xgboost
  - tabular-classification
  - bmw
  - car-model-prediction
  - automotive
datasets:
  - custom
inference: false
model-index:
  - name: BMW Model Predictor
    results: []
---

# 🚗 BMW Model Predictor

This machine learning model predicts the **BMW car model** using **XGBoost** based on the following features:

- `engine_power` (horsepower)
- `price` (EUR)
- `mileage` (km)
- `fuel` type
- `car_type` (e.g., sedan, SUV)

## 🔧 Training Information

- Model: XGBoost Classifier
- Training set: Custom dataset from `bmw_pricing_challenge.csv`
- Target variable: `model_key` (encoded)
- Preprocessing: Label Encoding for categorical features (`fuel`, `car_type`, `model_key`)
- Saved with: `joblib`

## 📁 Files

-Load encoders from .npy
fuel_encoder = LabelEncoder(); fuel_encoder.classes_ = np.load("fuel_classes.npy", allow_pickle=True)
car_type_encoder = LabelEncoder(); car_type_encoder.classes_ = np.load("car_type_classes.npy", allow_pickle=True)
target_encoder = LabelEncoder(); target_encoder.classes_ = np.load("target_classes.npy", allow_pickle=True)
- 
## 📊 BMW Model Predictor Overview

![BMW Model Predictor](./bmw-model-predictor.png)


## 🚀 How to Use

import sys, subprocess, os, json
def pip_install(pkg):
    try:
        __import__(pkg)
    except ImportError:
        subprocess.check_call([sys.executable, "-m", "pip", "install", pkg])

for p in ["huggingface_hub", "numpy", "scikit_learn", "xgboost"]:
    pip_install(p)

from huggingface_hub import snapshot_download
import numpy as np
from sklearn.preprocessing import LabelEncoder
from xgboost import XGBClassifier

# 1) Download model files
repo_id = "MahmutCanBoran/bmw-model-predictor"
local_dir = "bmw-model-predictor"
snapshot_download(repo_id=repo_id, repo_type="model", local_dir=local_dir, allow_patterns=["*"])

# 2) Path and file controls
path = local_dir + os.sep
required_files = ["bmw_model.json", "fuel_classes.npy", "car_type_classes.npy", "target_classes.npy"]
missing = [f for f in required_files if not os.path.exists(path + f)]
if missing:
    raise FileNotFoundError(f"Şu dosyalar eksik: {missing}\nKlasör içeriği: {os.listdir(path)}")

# 3) Restore the encoders.
fuel_le = LabelEncoder();     fuel_le.classes_     = np.load(path + "fuel_classes.npy", allow_pickle=True)
car_type_le = LabelEncoder(); car_type_le.classes_ = np.load(path + "car_type_classes.npy", allow_pickle=True)
target_le = LabelEncoder();   target_le.classes_   = np.load(path + "target_classes.npy", allow_pickle=True)

# 4) Upload XGBoost model(JSON)
model = XGBClassifier()
model.load_model(path + "bmw_model.json")

# 5)Helper: encode the input and make a prediction
def predict_bmw_model(engine_power:int, mileage:int, price:float, fuel:str, car_type:str):
    # kategori doğrulama (hata mesajları daha anlaşılır olsun)
    if fuel not in set(fuel_le.classes_):
        raise ValueError(f"fuel='{fuel}' geçersiz. Geçerli fuel sınıfları: {list(fuel_le.classes_)}")
    if car_type not in set(car_type_le.classes_):
        raise ValueError(f"car_type='{car_type}' geçersiz. Geçerli car_type sınıfları: {list(car_type_le.classes_)}")

    X = np.array([[
        engine_power,
        mileage,
        price,
        fuel_le.transform([fuel])[0],
        car_type_le.transform([car_type])[0]
    ]], dtype=float)

    y_pred = model.predict(X)
    return target_le.inverse_transform(y_pred)[0]

# 6) Info: display the available classes
print("Fuel classes:", list(fuel_le.classes_))
print("Car type classes:", list(car_type_le.classes_))

# 7) Example prediction
example = {"engine_power": 200, "mileage": 50000, "price": 25000, "fuel": "diesel", "car_type": "suv"}
pred = predict_bmw_model(**example)
print("Predicted BMW model:", pred)