Update app.py

app.py

@@ -4,7 +4,7 @@ import tempfile
 import os
 from datetime import datetime
 import numpy as np
-import MLForecast
+import matplotlib.pyplot as plt
 
 from statsforecast import StatsForecast
 from statsforecast.models import (
@@ -28,7 +28,41 @@ from utilsforecast.losses import *
 from mlforecast import MLForecast
 from lightgbm import LGBMRegressor
 
-#Function to generate and return a plot for validation results
+# Foundation Models
+try:
+    from chronos import ChronosPipeline
+    import torch
+    CHRONOS_AVAILABLE = True
+except:
+    CHRONOS_AVAILABLE = False
+
+try:
+    from uni2ts.model.moirai import MoiraiForecast
+    MOIRAI_AVAILABLE = True
+except:
+    MOIRAI_AVAILABLE = False
+
+# Helper function to calculate date offset based on frequency and horizon
+def calculate_date_offset(freq, horizon):
+    """Calculate a timedelta based on frequency code and horizon"""
+    if freq == 'H':
+        return pd.Timedelta(hours=horizon)
+    elif freq == 'D':
+        return pd.Timedelta(days=horizon)
+    elif freq == 'B':
+        return pd.Timedelta(days=int(horizon * 1.4))
+    elif freq == 'WS':
+        return pd.Timedelta(weeks=horizon)
+    elif freq == 'MS':
+        return pd.Timedelta(days=horizon * 30)
+    elif freq == 'QS':
+        return pd.Timedelta(days=horizon * 90)
+    elif freq == 'YS':
+        return pd.Timedelta(days=horizon * 365)
+    else:
+        return pd.Timedelta(days=horizon)
+
+# Function to generate and return a plot for validation results
 def create_forecast_plot(forecast_df, original_df, title="Forecasting Results", horizon=None, freq='D'):
     plt.figure(figsize=(12, 7))
     unique_ids = forecast_df['unique_id'].unique()
@@ -88,22 +122,6 @@ def create_forecast_plot(forecast_df, original_df, title="Forecasting Results",
 
     return fig
 
-     
-
-# Foundation Models
-try:
-    from chronos import ChronosPipeline
-    import torch
-    CHRONOS_AVAILABLE = True
-except:
-    CHRONOS_AVAILABLE = False
-
-try:
-    from uni2ts.model.moirai import MoiraiForecast
-    MOIRAI_AVAILABLE = True
-except:
-    MOIRAI_AVAILABLE = False
-
 # Function to load and process uploaded CSV
 def load_data(file):
     if file is None:
@@ -125,28 +143,6 @@ def load_data(file):
         return df, "Data loaded successfully!"
     except Exception as e:
         return None, f"Error loading data: {str(e)}"
-     
-
-# Helper function to calculate date offset based on frequency and horizon
-def calculate_date_offset(freq, horizon):
-    """Calculate a timedelta based on frequency code and horizon"""
-    if freq == 'H':
-        return pd.Timedelta(hours=horizon)
-    elif freq == 'D':
-        return pd.Timedelta(days=horizon)
-    elif freq == 'B':
-        return pd.Timedelta(days=int(horizon * 1.4))
-    elif freq == 'WS':
-        return pd.Timedelta(weeks=horizon)
-    elif freq == 'MS':
-        return pd.Timedelta(days=horizon * 30)
-    elif freq == 'QS':
-        return pd.Timedelta(days=horizon * 90)
-    elif freq == 'YS':
-        return pd.Timedelta(days=horizon * 365)
-    else:
-        return pd.Timedelta(days=horizon)
-     
 
 # Main forecasting function
 def run_forecast(
@@ -170,9 +166,6 @@ def run_forecast(
         # Prepare data - only required columns for models without predictors
         df_basic = df[['unique_id', 'ds', 'y']].copy()
 
-        # For models that need predictors, prepare full feature set
-        # (This would be expanded based on your feature engineering)
-
         # Initialize models list
         models = []
         models_need_predictors = []
@@ -220,7 +213,7 @@ def run_forecast(
             if models_need_predictors:
                 sf_pred = StatsForecast(models=models_need_predictors, freq=frequency, n_jobs=-1)
                 cv_df_pred = sf_pred.cross_validation(
-                    df=df_basic,  # Use df with predictors when implemented
+                    df=df_basic,
                     h=int(h),
                     step_size=int(step_size),
                     n_windows=int(num_windows)
@@ -235,7 +228,6 @@ def run_forecast(
                 return None, None, None, None, None, [], "No models selected"
 
         else:  # Fixed Window
-            # Similar logic for fixed window
             # Split data
             train_df = []
             for uid in df_basic['unique_id'].unique():
@@ -396,8 +388,6 @@ def run_forecast(
         error_msg = f"Error: {str(e)}\n\n{traceback.format_exc()}"
         return None, None, None, None, None, [], error_msg
 
-     
-
 # Gradio Interface
 with gr.Blocks(title="Duke Energy Forecasting App") as app:
     gr.Markdown("""