Create app.py

app.py

@@ -0,0 +1,73 @@
+# app.py
+
+import joblib
+import pandas as pd
+from pydantic import BaseModel
+from fastapi import FastAPI, HTTPException
+from huggingface_hub import hf_hub_download 
+
+# --- KONFIGURASI HF HUB ---
+HF_REPO_ID = "aephidayatuloh/bank-model" 
+HF_MODEL_FILENAME = "random_forest_bank_marketing_pipeline.joblib"
+
+# --- DEFINISI DATA INPUT (Pydantic) ---
+# [Pastikan ini sama persis dengan yang Anda gunakan sebelumnya]
+class PredictionInput(BaseModel):
+    age: int
+    balance: int
+    day: int
+    campaign: int
+    pdays: int
+    previous: int
+    job: str
+    marital: str
+    education: str
+    default: str
+    housing: str
+    loan: str
+    contact: str
+    month: str
+    poutcome: str
+
+# --- SETUP MODEL (DIJALANKAN SEKALI SAAT STARTUP) ---
+app = FastAPI(title="Bank Deposit Prediction (Docker)")
+
+@app.on_event("startup")
+def load_model():
+    global MODEL_PIPELINE
+    try:
+        # Download model dari HF Hub (direkomendasikan)
+        downloaded_model_path = hf_hub_download(
+            repo_id=HF_REPO_ID, 
+            filename=HF_MODEL_FILENAME
+        )
+        MODEL_PIPELINE = joblib.load(downloaded_model_path)
+        print("✅ Model berhasil dimuat dari Hugging Face Hub.")
+    except Exception as e:
+        print(f"❌ Gagal memuat model: {e}")
+        MODEL_PIPELINE = None
+
+
+# --- ENDPOINT PREDIKSI ---
+
+@app.get("/")
+def home():
+    return {"status": "ok", "message": "FastAPI is running inside Docker on HF Spaces."}
+
+@app.post("/predict")
+def predict(data: PredictionInput):
+    if MODEL_PIPELINE is None:
+        raise HTTPException(status_code=500, detail="Model gagal dimuat.")
+
+    try:
+        input_df = pd.DataFrame([data.dict()])
+        
+        prediction = MODEL_PIPELINE.predict(input_df)[0]
+        prediction_proba = MODEL_PIPELINE.predict_proba(input_df)[0].tolist()
+
+        return {
+            "prediction_class": int(prediction),
+            "probability": prediction_proba
+        }
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Prediction error: {e}")