Update app.py

app.py

@@ -1,82 +1,106 @@
 import gradio as gr
 import pandas as pd
-import numpy as np
-from core.data import preprocess_data
-from core.trainer import train_model
+from core.data import get_data
+from core.model_runner import run_model
 from core.plot import plot_forecast, plot_metrics
-from core.utils import plot_loss_curve
-from core.model_runner import get_model
-
-def forecast_pipeline(data, model_type, window_size, forecast_steps, epochs):
-    try:
-        df = pd.read_csv(data.name) if hasattr(data, 'name') else pd.read_csv("sample_data.csv")
-        df = df.dropna()
-
-        X_train, y_train, X_test, y_test, scaler = preprocess_data(df, window_size, forecast_steps)
-
-        model = get_model(model_type, input_shape=X_train.shape[1:])
-        history, predictions = train_model(model, X_train, y_train, X_test, y_test, epochs)
-
-        # Inverse transform predictions and actuals
-        predicted = scaler.inverse_transform(predictions)
-        actual = scaler.inverse_transform(y_test)
-
-        # Metrics
-        mse = np.mean((predicted - actual)**2)
-        mae = np.mean(np.abs(predicted - actual))
-        r2 = 1 - (np.sum((predicted - actual) ** 2) / np.sum((actual - np.mean(actual)) ** 2))
-
-        result = {
-            "actual": actual.flatten(),
-            "predicted": predicted.flatten(),
-            "loss": history.history['loss'],
-            "val_loss": history.history.get('val_loss', []),
-            "metrics": {
-                "MSE": mse,
-                "MAE": mae,
-                "R²": r2
-            }
-        }
-
-        return (
-            plot_forecast(result),
-            plot_loss_curve(result),
-            plot_metrics(result),
-            result["metrics"]
-        )
-    except Exception as e:
-        return f"Error: {str(e)}", None, None, None
-
-# Gradio interface
-with gr.Blocks(title="📈 MarketPredict: Forecasting Dashboard") as demo:
-    gr.Markdown("# 🧠 MarketPredict: Advanced Time Series Forecasting")
-    gr.Markdown("Upload a CSV with time-series data or use the sample. Select model and parameters to begin training and see predictions.")
-
-    with gr.Row():
-        with gr.Column():
-            file_input = gr.File(label="Upload CSV", file_types=[".csv"])
-            model_choice = gr.Dropdown(
-                choices=["LSTM", "GRU", "CNN", "Transformer", "MLP"],
-                label="Select Model",
-                value="LSTM"
+from config import AVAILABLE_MODELS, DEFAULT_TICKERS, DEFAULT_PARAMS
+
+
+def main_interface():
+    with gr.Blocks(theme=gr.themes.Soft()) as app:
+        gr.Markdown("# 📈 AI Forecasting Studio")
+        gr.Markdown("Upload time series data or fetch from Yahoo Finance and forecast with deep learning models (LSTM, GRU, Transformer, etc).")
+
+        with gr.Row():
+            with gr.Column(scale=1):
+                data_src = gr.Radio(["Yahoo Finance", "Upload CSV"], label="Data Source", value="Yahoo Finance")
+
+                ticker = gr.Dropdown(choices=DEFAULT_TICKERS, label="Ticker Symbol", value="BTC-USD", visible=True)
+                file_upload = gr.File(label="Upload CSV", visible=False, file_types=[".csv"])
+
+                start_date = gr.Textbox(label="Start Date (YYYY-MM-DD)", value="2022-01-01")
+                end_date = gr.Textbox(label="End Date (YYYY-MM-DD)", value="2023-12-31")
+                horizon = gr.Slider(minimum=10, maximum=200, step=1, label="Forecast Horizon (Days)", value=30)
+
+                gr.Markdown("## ⚙️ Model Configuration")
+
+                model = gr.Dropdown(choices=AVAILABLE_MODELS, label="Model", value="LSTM")
+                hidden_units = gr.Slider(minimum=8, maximum=512, step=8, label="Hidden Units", value=64)
+                n_layers = gr.Slider(minimum=1, maximum=5, step=1, label="Number of Layers", value=2)
+                epochs = gr.Slider(minimum=10, maximum=300, step=10, label="Epochs", value=100)
+                learning_rate = gr.Slider(minimum=1e-5, maximum=0.01, step=1e-5, label="Learning Rate", value=0.001)
+                window_size = gr.Slider(minimum=5, maximum=90, step=1, label="Window Size", value=30)
+                test_split = gr.Slider(minimum=0.05, maximum=0.5, step=0.01, label="Test Split (Fraction)", value=0.2)
+
+                run_btn = gr.Button("🚀 Train & Predict")
+                status = gr.Textbox(label="Status", lines=3, interactive=False)
+
+            with gr.Column(scale=2):
+                forecast_plot = gr.Plot(label="📊 Forecast Results")
+                error_plot = gr.Plot(label="📉 Backtest / Error Analysis")
+
+        # === Backend Logic ===
+
+        def run_pipeline(data_src, ticker, file_upload, start_date, end_date, horizon,
+                         model, hidden_units, n_layers, epochs, learning_rate,
+                         window_size, test_split):
+
+            try:
+                start_dt = pd.to_datetime(start_date)
+                end_dt = pd.to_datetime(end_date)
+
+                if start_dt >= end_dt:
+                    return None, None, "❌ End date must be after start date."
+
+                df = get_data(data_src, ticker, file_upload, start_date, end_date)
+                if df is None or df.empty:
+                    return None, None, "❌ Failed to load or parse dataset."
+
+                result = run_model(
+                    df=df,
+                    model_name=model,
+                    horizon=horizon,
+                    hidden_units=hidden_units,
+                    n_layers=n_layers,
+                    epochs=epochs,
+                    learning_rate=learning_rate,
+                    window_size=window_size,
+                    test_split=test_split
+                )
+
+                forecast_fig = plot_forecast(result)
+                error_fig = plot_metrics(result)
+
+                return forecast_fig, error_fig, "✅ Forecast complete!"
+            except Exception as e:
+                return None, None, f"❌ Error occurred: {str(e)}"
+
+        def toggle_data_input(src):
+            return (
+                gr.update(visible=(src == "Yahoo Finance")),
+                gr.update(visible=(src == "Upload CSV"))
             )
-            window_slider = gr.Slider(5, 60, value=30, step=5, label="Window Size")
-            forecast_slider = gr.Slider(1, 20, value=5, step=1, label="Forecast Steps")
-            epoch_slider = gr.Slider(5, 200, value=50, step=5, label="Epochs")
-
-            run_btn = gr.Button("🚀 Run Forecast")
-
-        with gr.Column():
-            forecast_plot = gr.Plot(label="📉 Forecast vs Actual")
-            loss_plot = gr.Plot(label="📉 Training Loss Curve")
-            metrics_plot = gr.Plot(label="📊 Error Metrics")
-            metrics_out = gr.JSON(label="📌 Evaluation Metrics")
-
-    run_btn.click(
-        fn=forecast_pipeline,
-        inputs=[file_input, model_choice, window_slider, forecast_slider, epoch_slider],
-        outputs=[forecast_plot, loss_plot, metrics_plot, metrics_out]
-    )
-
-if __name__ == "__main__":
-    demo.launch()
+
+        # === Event Bindings ===
+        run_btn.click(
+            fn=run_pipeline,
+            inputs=[
+                data_src, ticker, file_upload,
+                start_date, end_date, horizon,
+                model, hidden_units, n_layers, epochs,
+                learning_rate, window_size, test_split
+            ],
+            outputs=[forecast_plot, error_plot, status]
+        )
+
+        data_src.change(
+            fn=toggle_data_input,
+            inputs=[data_src],
+            outputs=[ticker, file_upload]
+        )
+
+    return app
+
+
+if __name__ == '__main__':
+    main_interface().launch()