Update app.py

app.py

@@ -1,54 +1,83 @@
 import pandas as pd
 import gradio as gr
-from prophet import Prophet
+import lightgbm as lgb
 import plotly.graph_objects as go
+import numpy as np
+from sklearn.model_selection import train_test_split
 
 # Load dataset
-DATA_PATH = "/content/AAPL_stock_data_finalversion (1).csv"
+DATA_PATH = "/mnt/data/AAPL_stock_data_finalversion (1).csv"
 data = pd.read_csv(DATA_PATH)
 
 # Ensure the dataset has the required columns
 data['Date'] = pd.to_datetime(data['Date'])
-
-# Rename columns for Prophet compatibility
-data = data.rename(columns={'Date': 'ds', 'Close': 'y'})
+data['DateNumeric'] = data['Date'].map(pd.Timestamp.toordinal)
 
 # Define prediction function
 def stock_analysis(start_date, end_date):
     try:
         # Filter data by date range
-        filtered_data = data[(data['ds'] >= pd.to_datetime(start_date)) & (data['ds'] <= pd.to_datetime(end_date))]
+        filtered_data = data[(data['Date'] >= pd.to_datetime(start_date)) & (data['Date'] <= pd.to_datetime(end_date))]
 
         if filtered_data.empty:
             return "No data available for the given date range.", None, None
 
-        # Fit the Prophet model
-        model = Prophet()
-        model.fit(filtered_data)
+        # Prepare data for forecasting
+        X = filtered_data[['DateNumeric']]
+        y = filtered_data['Close']
 
-        # Predict future stock prices (next 30 days)
-        future_dates = model.make_future_dataframe(periods=30)
-        forecast = model.predict(future_dates)
+        # Train-test split
+        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
+
+        # Prepare LightGBM dataset
+        train_data = lgb.Dataset(X_train, label=y_train)
 
-        # Filter only the forecasted dates
-        forecast = forecast[['ds', 'yhat']].tail(30).rename(columns={'ds': 'Date', 'yhat': 'Predicted Close Price'})
+        # LightGBM parameters
+        params = {
+            'objective': 'regression',
+            'metric': 'rmse',
+            'boosting_type': 'gbdt',
+            'num_leaves': 31,
+            'learning_rate': 0.05,
+            'verbose': -1
+        }
+
+        # Train LightGBM model
+        model = lgb.train(params, train_data, num_boost_round=100)
+
+        # Predict future stock prices (next 30 days)
+        last_date = filtered_data['Date'].iloc[-1]
+        future_dates = pd.date_range(last_date, periods=30, freq='B')
+        future_dates_numeric = future_dates.map(pd.Timestamp.toordinal).values.reshape(-1, 1)
+        future_predictions = model.predict(future_dates_numeric)
+
+        # Add trend or variation to predictions based on historical data
+        historical_trend = np.gradient(filtered_data['Close'].values)
+        trend_mean = np.mean(historical_trend) if len(historical_trend) > 0 else 0
+        future_predictions = future_predictions + np.linspace(0, trend_mean * 30, len(future_predictions))
+
+        # Create a DataFrame for predictions
+        future_df = pd.DataFrame({
+            'Date': future_dates,
+            'Predicted Close Price': future_predictions.flatten()
+        })
 
         # Create a plot using Plotly
         fig = go.Figure()
 
         # Add historical data to the plot
         fig.add_trace(go.Scatter(
-            x=filtered_data['ds'],
-            y=filtered_data['y'],
-            mode='lines',
+            x=filtered_data['Date'], 
+            y=filtered_data['Close'], 
+            mode='lines', 
             name='Historical Close Prices'
         ))
 
         # Add predicted data to the plot
         fig.add_trace(go.Scatter(
-            x=forecast['Date'],
-            y=forecast['Predicted Close Price'],
-            mode='lines',
+            x=future_df['Date'], 
+            y=future_df['Predicted Close Price'], 
+            mode='lines', 
             name='Predicted Close Prices'
         ))
 
@@ -60,7 +89,7 @@ def stock_analysis(start_date, end_date):
             template="plotly_white"
         )
 
-        return "Analysis completed!", fig, forecast
+        return "Analysis completed!", fig, future_df
 
     except Exception as e:
         return f"An error occurred: {str(e)}", None, None