added best model

app.py

@@ -1,7 +1,7 @@
 import uvicorn
 import joblib
 import numpy as np
-import tensorflow as tf
+import xgboost as xgb
 import sys
 import os
 from fastapi import FastAPI, HTTPException
@@ -9,7 +9,7 @@ from pydantic import BaseModel
 from fastapi.middleware.cors import CORSMiddleware
 
 # --- FILE PATHS (MUST match the export paths from your training script) ---
-MODEL_EXPORT_PATH = "final.keras"
+MODEL_EXPORT_PATH = "best_model_xgboost.h5"
 SCALER_EXPORT_PATH = "scaler.joblib"
 # -------------------------------------------------------------------------
 
@@ -25,23 +25,21 @@ class ConcreteInput(BaseModel):
 
 # --- Initialize FastAPI App ---
 app = FastAPI(
-    title="Concrete Strength Predictor",
-    description="API for predicting concrete compressive strength using a Keras ANN model."
+    title="Concrete Strength Predictor (XGBoost)",
+    description="API for predicting concrete compressive strength using XGBoost model."
 )
 
 # --- CORS Configuration ---
-origins = ["*"] 
+origins = ["*"]
 
 app.add_middleware(
     CORSMiddleware,
     allow_origins=origins,
     allow_credentials=True,
-    allow_methods=["*"],  # Allows all methods (POST, GET, etc.)
-    allow_headers=["*"],  # Allows all headers
+    allow_methods=["*"],
+    allow_headers=["*"],
 )
 
-
-
 # --- Global variables to store the loaded model and scaler ---
 model = None
 scaler = None
@@ -49,10 +47,10 @@ scaler = None
 # --- Startup Event Handler to Load Assets ---
 @app.on_event("startup")
 def load_assets():
-    """Load the Keras model and StandardScaler on application startup."""
+    """Load the XGBoost model and StandardScaler on application startup."""
     global model, scaler
-    print("Attempting to load model and scaler...")
-    
+    print("Attempting to load XGBoost model and scaler...")
+
     # Load Scaler
     try:
         scaler = joblib.load(SCALER_EXPORT_PATH)
@@ -64,16 +62,16 @@ def load_assets():
         print(f"❌ FATAL ERROR: Failed to load scaler: {e}")
         sys.exit(1)
 
-    # Load Model
+    # Load XGBoost Model
     try:
-        # Load the model outside the prediction function for performance
-        model = tf.keras.models.load_model(MODEL_EXPORT_PATH)
-        print(f"✅ Model loaded successfully from {MODEL_EXPORT_PATH}")
+        model = xgb.XGBRegressor()
+        model.load_model(MODEL_EXPORT_PATH)
+        print(f"✅ XGBoost model loaded successfully from {MODEL_EXPORT_PATH}")
     except FileNotFoundError:
         print(f"❌ FATAL ERROR: Model file not found at {MODEL_EXPORT_PATH}. Run the training script first!")
         sys.exit(1)
     except Exception as e:
-        print(f"❌ FATAL ERROR: Failed to load Keras model: {e}")
+        print(f"❌ FATAL ERROR: Failed to load XGBoost model: {e}")
         sys.exit(1)
 
 # --- Health Check Endpoint ---
@@ -85,37 +83,30 @@ def read_root():
 # --- Prediction Endpoint ---
 @app.post("/con-predict")
 async def predict_strength(data: ConcreteInput):
-    """
-    Accepts concrete component values and returns the predicted strength (MPa).
-    """
+    """Accepts concrete component values and returns the predicted strength (MPa)."""
     if model is None or scaler is None:
         raise HTTPException(status_code=503, detail="Model assets failed to load on startup.")
 
-    # 1. Convert Pydantic data object to a list in the correct order
-    # This list must match the feature order: Cement, Slag, Ash, Water, Superplasticizer, Coarse Aggregate, Fine Aggregate, Age
+    # Prepare input
     input_list = [
-        data.cement, data.slag, data.ash, data.water, 
-        data.superplasticizer, data.coarse_aggregate, data.fine_aggregate, data.age
+        data.cement, data.slag, data.ash, data.water,
+        data.superplasticizer, data.coarse_aggregate,
+        data.fine_aggregate, data.age
     ]
-    
-    # Convert list to NumPy array, reshaping to (1, 8) for a single sample
     input_array = np.array([input_list])
 
-    # 2. Scale the input data
+    # Scale input
     input_scaled = scaler.transform(input_array)
 
-    # 3. Predict the strength
+    # Predict strength
     prediction = model.predict(input_scaled)
-    
-    # Extract the single prediction value
-    predicted_strength = float(prediction[0][0])
+    predicted_strength = float(prediction[0])
 
-    # 4. Return the result
+    # Return result
     return {
         "predicted_strength_mpa": round(predicted_strength, 2),
         "input_features": data.dict()
     }
 
 if __name__ == "__main__":
-    # Use --reload to automatically restart the server on file changes
     uvicorn.run("app:app", host="0.0.0.0", port=7860)

best_model_xgboost.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1abba2c0ce8f2a746c9f0ad2033de0c2595efdf00a68d86466c8b1ada97dfca7
+size 918884