Researching the Efficiency of Machine Learning Methods Used in Forecasting Demand for Certain Types of Goods

The article deals with the problem of improving the efficiency of forecasting demand for certain types of goods. The goal of the study is to determine the most effective machine learning method for this forecasting. The relevance of the work is due to the fact that traditional statistical methods for forecasting demand do not provide the ability to process effectively the currently received volumes of information on sales and trade; besides, not all machine learning methods for predicting numerical values are suitable for the subject area under consideration. The object of the study is the arising demand for certain types of goods. The subject of the study is the effectiveness indicators (or metrics) of machine learning methods used to forecast demand for certain types of goods. When analyzing the subject area, mean absolute error, the root mean square error and coefficient of determination were selected as metrics used to evaluate the effectiveness of a particular method. As a result, the most suitable algorithms for the most effective forecast of demand for various types of goods were identified. Finally, conclusions are drawn.