Update app.py

app.py

@@ -55,13 +55,32 @@ def calculate_metrics(actual, predicted):
 
 # Function to plot ACF and PACF
 def plot_acf_pacf(series):
-    fig, ax = plt.subplots(1, 2, figsize=(12, 6))
-    plot_acf(series, lags=20, ax=ax[0])
-    plot_pacf(series, lags=20, ax=ax[1])
-    ax[0].set_title('Auto-Correlation Function (ACF)')
-    ax[1].set_title('Partial Auto-Correlation Function (PACF)')
-    st.pyplot(fig)
+    try:
+        # Ensure the series has enough data points for the desired number of lags
+        if len(series) < 20:
+            st.write("The series is too short to generate ACF and PACF plots with 20 lags.")
+            return  # Skip plotting
 
+        fig, ax = plt.subplots(1, 2, figsize=(12, 6))
+        
+        # Plot ACF and PACF with error handling for shape mismatch
+        plot_acf(series, lags=20, ax=ax[0])
+        plot_pacf(series, lags=20, ax=ax[1])
+        
+        # Set titles for the plots
+        ax[0].set_title('Auto-Correlation Function (ACF)')
+        ax[1].set_title('Partial Auto-Correlation Function (PACF)')
+
+        # Display the plots using Streamlit
+        st.pyplot(fig)
+        
+    except ValueError as e:
+        st.error(f"An error occurred while plotting ACF and PACF: {e}")
+    
+def calculate_metrics2(actual, predicted):
+    mae = mean_absolute_error(actual, predicted)
+    rmse = np.sqrt(mean_squared_error(actual, predicted))
+    return mae, rmse
 # Forecasting button
 if st.button("Forecast"):
     # Prepare the historical data for comparison
@@ -149,20 +168,7 @@ if st.button("Forecast"):
         ax.set_ylabel('Sales Quantity')
         ax.legend()
         st.pyplot(fig)
-import matplotlib.pyplot as plt
-import numpy as np
-import pandas as pd
-import pickle
-import os
-import streamlit as st
-from sklearn.metrics import mean_absolute_error, mean_squared_error
-from statsmodels.tsa.arima.model import ARIMA
 
-# Function to calculate evaluation metrics
-def calculate_metrics(actual, predicted):
-    mae = mean_absolute_error(actual, predicted)
-    rmse = np.sqrt(mean_squared_error(actual, predicted))
-    return mae, rmse
 
 if st.button("Compare Models"):
     # Load and forecast with ARIMA
@@ -204,7 +210,7 @@ if st.button("Compare Models"):
         axs[0, 0].legend()
 
         # Calculate and display ARIMA metrics
-        arima_mae, arima_rmse = calculate_metrics(product_sales_series[-15:], st.session_state.arima_forecast)
+        arima_mae, arima_rmse = calculate_metrics2(product_sales_series[-15:], st.session_state.arima_forecast)
         st.write(f'**ARIMA Model Metrics:**\n- MAE: {arima_mae:.2f}\n- RMSE: {arima_rmse:.2f}')
 
     # Plot Decision Tree forecast
@@ -216,7 +222,7 @@ if st.button("Compare Models"):
         axs[0, 1].legend()
 
         # Calculate and display Decision Tree metrics
-        dt_mae, dt_rmse = calculate_metrics(product_sales_series[-15:], st.session_state.dt_forecast)
+        dt_mae, dt_rmse = calculate_metrics2(product_sales_series[-15:], st.session_state.dt_forecast)
         st.write(f'**Decision Tree Model Metrics:**\n- MAE: {dt_mae:.2f}\n- RMSE: {dt_rmse:.2f}')
 
     # Plot XGBoost forecast
@@ -228,7 +234,7 @@ if st.button("Compare Models"):
         axs[1, 0].legend()
 
         # Calculate and display XGBoost metrics
-        xgb_mae, xgb_rmse = calculate_metrics(product_sales_series[-15:], st.session_state.xgb_forecast)
+        xgb_mae, xgb_rmse = calculate_metrics2(product_sales_series[-15:], st.session_state.xgb_forecast)
         st.write(f'**XGBoost Model Metrics:**\n- MAE: {xgb_mae:.2f}\n- RMSE: {xgb_rmse:.2f}')
 
     # Hide the last subplot (bottom right) if not needed