Update app.py

app.py

@@ -4,22 +4,24 @@ from typing import Optional
 import pandas as pd
 import joblib
 import os
-from sklearn.metrics import classification_report
-from sklearn.model_selection import train_test_split
 
+# Initialize FastAPI app
 app = FastAPI()
 
 # --- Model paths ---
 TFIDF_VECTORIZER_PATH = "models/tfidf_vectorizer.pkl"
-MODELS_PATH = "models/xgb_model.pkl"
+MODELS_PATH = "models/xgb_models.pkl"
 LABEL_ENCODERS_PATH = "models/label_encoders.pkl"
 
-# --- Load models ---
-tfidf_vectorizer = joblib.load(TFIDF_VECTORIZER_PATH)
-models = joblib.load(MODELS_PATH)
-label_encoders = joblib.load(LABEL_ENCODERS_PATH)
+# --- Load Models ---
+try:
+    tfidf_vectorizer = joblib.load(TFIDF_VECTORIZER_PATH)
+    models = joblib.load(MODELS_PATH)
+    label_encoders = joblib.load(LABEL_ENCODERS_PATH)
+except Exception as e:
+    raise RuntimeError(f"Model loading failed: {e}")
 
-# --- Input Schema ---
+# --- Input Schemas ---
 class TransactionData(BaseModel):
     Transaction_Id: str
     Hit_Seq: int
@@ -88,81 +90,46 @@ class TransactionData(BaseModel):
 class PredictionRequest(BaseModel):
     transaction_data: TransactionData
 
+# --- Root Route ---
 @app.get("/")
-async def root():
+def health_check():
     return {"status": "healthy", "message": "XGBoost TF-IDF API is running"}
 
+# --- Predict Route ---
 @app.post("/predict")
 async def predict(request: PredictionRequest):
     try:
         input_data = pd.DataFrame([request.transaction_data.dict()])
 
+        # Concatenate relevant fields into a single string
         text_input = "\n".join([
-            str(input_data[col].iloc[0])
-            for col in input_data.columns
-            if pd.notna(input_data[col].iloc[0])
+            str(input_data[col].iloc[0]) for col in input_data.columns if pd.notna(input_data[col].iloc[0])
         ])
 
+        # TF-IDF transform
         X_tfidf = tfidf_vectorizer.transform([text_input])
 
+        # Predict for each label
         response = {}
         for label, model in models.items():
             proba = model.predict_proba(X_tfidf)[0]
-            pred = proba.argmax()
-            decoded_label = label_encoders[label].inverse_transform([pred])[0]
+            pred_idx = proba.argmax()
+            pred_label = label_encoders[label].inverse_transform([pred_idx])[0]
             class_probs = {
                 label_encoders[label].classes_[i]: float(prob)
                 for i, prob in enumerate(proba)
             }
             response[label] = {
-                "prediction": decoded_label,
+                "prediction": pred_label,
                 "probabilities": class_probs
             }
 
         return response
 
     except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-@app.get("/validate")
-async def validate_model():
-    try:
-        DATA_PATH = "data.csv"  # Ensure this file is present
-        df = pd.read_csv(DATA_PATH)
-        df.dropna(subset=["Sanction_Context"] + list(models.keys()), inplace=True)
-
-        def concat_text(row):
-            return "\n".join([
-                str(row.get(col, "")) for col in row.index
-            ])
-
-        df["combined_text"] = df.apply(concat_text, axis=1)
-        X = tfidf_vectorizer.transform(df["combined_text"])
-
-        results = {}
-
-        for label in models:
-            encoder = label_encoders[label]
-            y = encoder.transform(df[label])
-            _, X_test, _, y_test = train_test_split(
-                X, y, test_size=0.2, random_state=42
-            )
-            model = models[label]
-            y_pred = model.predict(X_test)
-
-            report = classification_report(
-                encoder.inverse_transform(y_test),
-                encoder.inverse_transform(y_pred),
-                output_dict=True
-            )
-
-            results[label] = report
-
-        return {"validation_reports": results}
-
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=f"Validation failed: {e}")
+        raise HTTPException(status_code=500, detail=f"Inference error: {str(e)}")
 
+# --- Run Locally (optional) ---
 if __name__ == "__main__":
     import uvicorn
     port = int(os.environ.get("PORT", 7860))