Özet:

In today’s digital world, the purchases are increasing and the data has reached a very large scale. The most obvious of the concepts brought by the industry was the multi-dimensional curse. This is why companies have a lot of difficulty making their purchasing decisions. The failure to properly make a long or short-term sales forecast will lead to many problems such as customer dissatisfaction, loss of money, the need for raw materials. From supply chain elements to manufacturers, retailers, suppliers, and customers, many parties may suffer damage from the wrong or missing sales forecast. Machine learning, which is one of the innovations of the artificial intelligence age, is a field that can quickly respond to the problems brought by many engineering applications, as well as to the sales predictions. In this study, a machine learning from the end to the end of the project process was addressed. The steps of any machine learning project and the data approach size are introduced. A linear regression from machine learning algorithms in the application section, Ridge, Lasso, ElasticNet, K-en close neighbor and Rastgele Forest algorithms have developed a sales forecast model for a real data set using. The lowest error algorithm in the developed model was Random Forest algorithm.

Anahtar Kelimeler: