Leveraging Economic and Market Variables for Improved Tile Demand Forecasting using Machine Learning

The dynamic nature of the tile market requires precise demand forecasting to ensure inventory optimization, cost control, and enhanced customer satisfaction. This study focuses on developing a robust forecasting model that integrates time series and regression techniques to predict tile sales. Utilizing historical sales data, economic indicators such as GDP growth, inflation, and interest rates, and customer preferences, the research evaluates the performance of various machine learning models, including CatBoost, ARIMA, and Gradient Boosting. The results indicate that models integrating multiple variables significantly outperform those relying on individual factors, with Gradient Boosting and CatBoost demonstrate the highest accuracy for monthly and quarterly forecasts. These findings highlight the value of combining historical sales data, market trends, and economic indicators to achieve reliable sales predictions, thereby enabling more informed decision-making in the tile industry.