Sales Forecasting Demo

📄 Summary

This demo illustrates basic sales forecasting using JavaScript directly in the browser.

• Naive Forecast: Uses the last observed value for all future points.
• Moving Average Forecast: Computes the average of the previous 3 months to smooth short-term fluctuations.
• Linear Regression Forecast: Fits a straight line trend to the sales data to capture overall growth/decline.
• Quantile-Filtered Regression Forecast: Removes outliers using the interquartile range (IQR) before fitting a linear regression, making it more robust to anomalies.

The code calculates the MAPE (Mean Absolute Percentage Error) for each method and automatically identifies the best-performing forecasting approach.

Users can upload their own CSV (max 100 rows) and select different items to dynamically view forecasts and MAPE scores.