文章基本信息

标题：Benchmarking of Regression Algorithms and Time Series Analysis Techniques for Sales Forecasting
本地全文：下载
作者：Cagatay CATAL ; Kaan ECE ; Begum ARSLAN 等
期刊名称：Balkan Journal of Electrical & Computer Engineering
印刷版ISSN：2147-284X
出版年度：2019
卷号：7
期号：1
页码：20-26
DOI：10.17694/bajece.494920
语种：English
出版社：Kirklareli University
摘要：Predicting the sales amount as close as to theactual sales amount can provide many benefits to companies. Since the fashionindustry is not easily predictable, it is not straightforward to make anaccurate prediction of sales. In thisstudy, we applied not only regression methods in machine learning, but alsotime series analysis techniques to forecast the sales amount based on severalfeatures. We applied our models on Walmart sales data in Microsoft AzureMachine Learning Studio platform. The following regression techniques wereapplied: Linear Regression, Bayesian Regression, Neural Network Regression,Decision Forest Regression and Boosted Decision Tree Regression. In addition tothese regression techniques, the following time series analysis methods were implemented:Seasonal ARIMA, Non-Seasonal ARIMA, Seasonal ETS, Non -Seasonal ETS, NaiveMethod, Average Method and Drift Method. It was shown that Boosted Decision TreeRegression provides the best performance on this sales data. This project is apart of the development of a new decision support system for the retailindustry.
关键词：Sales forecasting;regression;machine learning;time series analysis