BFI Tech Test – Data Analytics Operation

API ini memprediksi kualitas wine berdasarkan fitur-fitur kimiawi menggunakan model machine learning yang telah dilatih.

---

🚀 Features

• 🔮 Prediksi kualitas wine (skor 0–10)
• ⚙️ Model ML berbasis RandomForestClassifier
• 🧪 Dibangun dengan FastAPI
• 🐳 Dikemas dalam Docker container
• ☁️ Deployment menggunakan Railway
• 📈 Monitoring dengan Prometheus & Grafana (di Play With Docker)

---

📁 Project Structure

.
├── app.py
├── train_model.py
├── predict.py
├── model.pkl
├── requirements.txt
├── Dockerfile
├── docker-compose.yml
├── prometheus.yml
├── curl_test.sh
├── README.md
└── WineQT.csv           

---

⚙️ Setup & Installation

1. 🔧 Model Training & API Deployment (via Railway)

1. Clone repositori ke Railway
2. Jalankan train_model.py untuk membuat model.pkl
3. Deploy FastAPI app dengan file app.py
4. Pastikan port 8000 aktif

2. 🧪 Test API Endpoint

Gunakan curl_test.sh atau langsung dengan curl:

sh curl_test.sh

Atau manual:

curl -X POST http://localhost:8000/predict \
-H "Content-Type: application/json" \
-d '{{
  "fixed_acidity": 7.4,
  "volatile_acidity": 0.7,
  "citric_acid": 0,
  "residual_sugar": 1.9,
  "chlorides": 0.076,
  "free_sulfur_dioxide": 11.0,
  "total_sulfur_dioxide": 34.0,
  "density": 0.9978,
  "pH": 3.51,
  "sulphates": 0.56,
  "alcohol": 9.4
}}'

---

📨 API Usage

Endpoint

POST /predict

Request Body

{{
  "fixed_acidity": 7.4,
  "volatile_acidity": 0.7,
  "citric_acid": 0,
  "residual_sugar": 1.9,
  "chlorides": 0.076,
  "free_sulfur_dioxide": 11.0,
  "total_sulfur_dioxide": 34.0,
  "density": 0.9978,
  "pH": 3.51,
  "sulphates": 0.56,
  "alcohol": 9.4
}}

Response

{{
  "prediction": 5
}}

---

3. 🔧 Monitoring & Logging (via PWD - Play With Docker)

1. Buka https://labs.play-with-docker.com/
2. Buat 3 terminal instance: FastAPI app, Prometheus, dan Grafana
3. Jalankan container dengan docker-compose up
4. Akses Prometheus di http://localhost:9090, Grafana di http://localhost:3000

---

📊 Monitoring Dashboard

Metode monitoring dilakukan dengan:

• Prometheus untuk scraping metrik dari endpoint /metrics
• Grafana untuk visualisasi
• Metrik:
  • api_request_count_total – Jumlah permintaan API
  • api_request_latency_seconds_bucket – Latensi request API

---

🧠 Model Information

• Model: Random Forest Classifier
• Dataset: Wine Quality (Red Wine)
• Fitur input: 11 fitur kimia (acidity, sugar, alcohol, dll)
• Output: Skor kualitas wine (0–10)

---

📦 Dependencies

• fastapi
• uvicorn
• scikit-learn
• pandas
• pydantic
• prometheus_client

Instalasi manual (jika tidak pakai Docker):

pip install -r requirements.txt

---

📬 Contact

Jika ada pertanyaan terkait proyek ini, silakan hubungi melalui email atau buat issue di repositori ini.