Update predict.py to auto-load/train model

src/predict.py

@@ -2,33 +2,43 @@
 
 import os
 import pickle
-import numpy as np
+import pandas as pd
+from model_training import train_model  # Your training function
 
-# Load trained model and label encoder
-model_path = os.path.join(os.path.dirname(__file__), '..', 'models', 'crop_model.pkl')
+# Path to the model file
+MODEL_PATH = os.path.join(os.path.dirname(__file__), '../models/crop_model.pkl')
 
-with open(model_path, 'rb') as f:
-    saved = pickle.load(f)
-
-model = saved['model']
-label_encoder = saved['label_encoder']
-
-def predict_crop(features):
+def load_model():
     """
-    Predict the best crop based on input features.
-
+    Loads the crop prediction model from disk.
+    If the model file does not exist, trains a new model and saves it.
+    """
+    if os.path.exists(MODEL_PATH):
+        # Load existing model
+        with open(MODEL_PATH, 'rb') as f:
+            model = pickle.load(f)
+    else:
+        # Train new model and save it
+        print("Model not found. Training a new model...")
+        model = train_model()  # Make sure your model_training.py has train_model() returning a fitted model
+        os.makedirs(os.path.dirname(MODEL_PATH), exist_ok=True)
+        with open(MODEL_PATH, 'wb') as f:
+            pickle.dump(model, f)
+        print("Model trained and saved.")
+    return model
+
+# Load model once when the app starts
+model = load_model()
+
+def predict_crop(input_features: pd.DataFrame):
+    """
+    Predicts the crop based on input features.
     Args:
-        features (list or array): [N, P, K, temperature, humidity, ph, rainfall]
-
+        input_features: pd.DataFrame with columns ['N', 'P', 'K', 'temperature', 'humidity', 'ph', 'rainfall']
     Returns:
-        str: Predicted crop name
+        List of predicted crop names
     """
-    features = np.array(features).reshape(1, -1)  # reshape for single sample
-    pred_encoded = model.predict(features)[0]     # get encoded label
-    crop_name = label_encoder.inverse_transform([pred_encoded])[0]  # decode label
-    return crop_name
+    predictions = model.predict(input_features)
+    return predictions.tolist()
+
 
-# Quick test
-if __name__ == "__main__":
-    test_features = [90, 42, 43, 20, 80, 6.5, 200]
-    print("Predicted Crop:", predict_crop(test_features))