Update app.py

app.py

@@ -4,7 +4,7 @@ import xgboost as xgb
 from sklearn.model_selection import train_test_split
 from sklearn.metrics import mean_squared_error
 import gradio as gr
-import json
+import os
 
 # API Endpoints
 WEATHER_API = "https://api.open-meteo.com/v1/forecast"
@@ -17,29 +17,30 @@ def fetch_weather_data():
         "latitude": 59.3293,  # Stockholm latitude
         "longitude": 18.0686,  # Stockholm longitude
         "hourly": ["temperature_2m", "precipitation", "wind_speed_10m", "humidity"],
+        "start": "2023-01-01",
+        "end": "2023-12-31"
     }
     response = requests.get(WEATHER_API, params=params)
-    return response.json()
+    response.raise_for_status()
+    return pd.DataFrame(response.json()["hourly"])
 
 # Fetch electricity price data
 def fetch_electricity_prices():
     response = requests.get(ELECTRICITY_PRICE_API + "/json/sv")
-    return response.json()
+    response.raise_for_status()
+    return pd.DataFrame(response.json())
 
 # Fetch energy production prices
 def fetch_energy_production_data():
-    # Customize parameters based on API documentation
+    # Placeholder for actual API parameters
     response = requests.get(ENERGY_PRODUCTION_API, params={})
-    return response.json()
+    response.raise_for_status()
+    return pd.DataFrame(response.json())
 
 # Prepare the dataset
 def prepare_dataset(weather_data, electricity_data, energy_data):
-    weather_df = pd.DataFrame(weather_data["hourly"])
-    electricity_df = pd.DataFrame(electricity_data)
-    energy_df = pd.DataFrame(energy_data)
-
-    # Combine dataframes
-    dataset = pd.concat([weather_df, electricity_df, energy_df], axis=1)
+    dataset = pd.concat([weather_data, electricity_data, energy_data], axis=1)
+    dataset = dataset.dropna()  # Remove any rows with missing values
     return dataset
 
 # Train the model
@@ -53,30 +54,31 @@ def train_model(dataset):
     predictions = model.predict(X_test)
 
     rmse = mean_squared_error(y_test, predictions, squared=False)
-    return model, rmse
+    model.save_model("electricity_price_model.json")
+    return rmse
 
-# Prediction function
-def predict_price(input_data):
-    # Convert input data to DataFrame for prediction
-    input_df = pd.DataFrame([input_data])
-    prediction = model.predict(input_df)
+# Load the model and make predictions
+def predict_price(features):
+    model = xgb.XGBRegressor()
+    model.load_model("electricity_price_model.json")
+    prediction = model.predict(pd.DataFrame([features]))
     return prediction[0]
 
-# Main function for Gradio interface
+# Gradio Interface
 def gradio_interface():
-    # Fetch data
-    weather_data = fetch_weather_data()
-    electricity_data = fetch_electricity_prices()
-    energy_data = fetch_energy_production_data()
-
-    # Prepare dataset and train model
-    dataset = prepare_dataset(weather_data, electricity_data, energy_data)
-    global model
-    model, rmse = train_model(dataset)
+    def wrapper(temp, precip, wind_speed, humidity, energy_price, electricity_price):
+        features = {
+            "temperature_2m": temp,
+            "precipitation": precip,
+            "wind_speed_10m": wind_speed,
+            "humidity": humidity,
+            "energy_price": energy_price,
+            "electricity_price": electricity_price
+        }
+        return predict_price(features)
 
-    # Define Gradio interface
     interface = gr.Interface(
-        fn=predict_price,
+        fn=wrapper,
         inputs=[
             gr.inputs.Number(label="Temperature (°C)"),
             gr.inputs.Number(label="Precipitation (mm)"),
@@ -86,7 +88,6 @@ def gradio_interface():
             gr.inputs.Number(label="Historical Electricity Price")
         ],
         outputs=gr.outputs.Textbox(label="Predicted Electricity Price"),
-        live=False,
         title="Electricity Price Prediction",
         description="Predict future electricity prices based on weather and energy data."
     )
@@ -94,4 +95,16 @@ def gradio_interface():
     interface.launch()
 
 if __name__ == "__main__":
+    # Fetch data
+    weather_data = fetch_weather_data()
+    electricity_data = fetch_electricity_prices()
+    energy_data = fetch_energy_production_data()
+
+    # Prepare dataset and train the model
+    dataset = prepare_dataset(weather_data, electricity_data, energy_data)
+    rmse = train_model(dataset)
+    print(f"Model trained with RMSE: {rmse}")
+
+    # Launch Gradio interface
     gradio_interface()
+