Sales Forecasting Web Application in Small and Medium Enterprise

Sales forecasting for small and medium enterprises (SMEs) is still challenging nowadays, especially in Indonesia, which was dominated by SMEs in their economic growth. On the other hand, many SMEs in Indonesia do not have the technology skills to leverage their business to be digitized. So in this paper, a web-based application in Puri Utami's as one of SMEs in Indonesia was built to help the owner predict the sales for the future. Two algorithms in the machine learning library for PHP (PHP-ML) were used as a library to create sales forecasting modules using Least Squares Linear Regression and Support Vector Regression (SVR) algorithm. The experimental results indicate that Least Squares Linear Regression achieves the best performances compared to SVR in both MSE and MAE evaluation metrics. In future works, a complex algorithm such as Deep Learning and enhance data preprocessing can be investigated further to improve the accuracy.