Comparative Analysis of Traditional and Modern Models for Forecasting Demand: Approaches and Features
Abstract
In inferential statistics, forecasting is a mathematical process by which the future value of one or more variables, such as demand, is estimated. The objective of this paper was to define the classification of the main types of forecasts. In addition, to propose some of the most representative models currently used to be implemented by small and medium-sized companies, those that, based on the literature consulted, have the greatest potential to achieve a successful demand forecast. It was found that forecasts could be univariate or multivariate; however, in order to find the alternatives that would require lower cost for the enterprise by data processing, only univariate time series forecast models were recommended, as they require only historical sales data of the company. Time series forecasting models were classified into three approaches: 1) statistical or traditional, of which the Holt-Winters or triple exponential smoothing model and the autoregressive integrated moving average (Arima) model were recommended, 2) machine learning, of which the random forest model and the large short-term memory (LSTM) recurrent neural network model stood out, 3) hybrids, of which the Arima-LSTM model and the Facebook Prophet model were suggested.
Downloads
This work is licensed under a Creative Commons Attribution 4.0 International License.
In order to promote the development and dissemination of research in education in Latin America, the Ibero-American Journal for Educational Research and Development (RIDE) adhered to the Budapest Open Access Initiative, which is why it is identified as a Open access publication. This means that any user can read the complete text of the articles, print them, download them, copy them, link them, distribute them and use the contents for other purposes. Creative Cummons licenses allow users to specify the rights to use an open access journal available on the Internet in such a way that users know the rules of publication. Authors who publish in this journal accept the following conditions: Authors they keep the author's rights and give the magazine the right of the first publication, with the work registered with the attribution license of Creative Commons, which allows third parties to use the published material whenever they mention the authorship of the work and the first publication in this The authors can make other independent and additional contractual agreements for the non-exclusive distribution of the version of the article published in this journal (eg, include it in an institutional repository or publish it in a book) as long as they clearly indicate that The work was published for the first time in this magazine. Authors are allowed and recommended to publish their work. low on the Internet (for example on institutional or personal pages) before and during the review and publication process, as it can lead to productive exchanges and to a greater and faster dissemination of the published work
