Multivariate forecast methodology with machine learning and integrated cloud-based reports

Authors

DOI:

https://doi.org/10.15665/rp.v18i2.2243

Abstract

Time series analysis is one of the most used tools to forecast based on past data. This work develops a multivariate methodology that attempts to overcome the difficulties of traditional time series analysis; utilizing new computational tools and data structures that facilitate integration with business applications and reduce the learning curve needed to obtain good forecasts. The methodology consists of five phases: (1) Importing the data directly from the cloud or from the user’s device, (2) Tidying and transforming, (3) Visualization, (4) Automatically model and validate the results, and (5) Communicate the obtained forecasts with an automated report. The methodology was used in an applied case considering ten time series from real retail sales indexes in Colombia, showing appreciable improvements with an average decrease on the Mean Absolute Percentage Error (MAPE) of 50.6%.

Author Biographies

  • Luis D. Chavarria-Munera, National University of Colombia

    Ingeniero Industrial. Departamento de Ingeniería de la Organización, Facultad de Minas, Universidad Nacional de Colombia. Medellín, Colombia

  • Juan M. Cogollo-Florez, Instituto Tecnológico Metropolitano

    Magíster en Ingeniería Administrativa. Profesor Asociado. Departamento de Calidad y Producción, Instituto Tecnológico Metropolitano – ITM. Medellín, Colombia

  • Alexander A. Correa-Espinal, National University of Colombia

    Doctor en Estadística e Investigación Operativa. Profesor Titular. Departamento de Ingeniería de la Organización, Facultad de Minas, Universidad Nacional de Colombia. Medellín, Colombia

Downloads

Published

2020-08-27

Issue

Vol. 18 No. 2 (2020): Julio - Diciembre

Section

Articles

License

Copyright (c) 2020 Juan M. Cogollo-Florez

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

The authors to publish in this journal agree to the following conditions:

  1. The authors transfer the copyright and give the the journal first publication right of the work registered with Creative Commons Attribution License, which allows third parties to use the published work on the condition of always mentioning the authorship and first publication in this journal.
  2. The authors may perform other independent and additional contractual arrangements for the non-exclusive distribution of the version of the article published in this issue (E.g., Inclusion in an institutional repository or publication in a book), it must be indicated clearly that the work was first published in this journal.
  3. It allows and encourages the authors to publish their work online (eg institutional or personal pages) before and during the review and publication process. It can lead to productive exchanges and greater and faster dissemination of the published work (see The Effect of Open Access)

Instructions to fill out Certificate of Originality and Copyright Assignment

  1. Click here and get the forms of Certificate of Originality and Copyright Assignment  .
  2. In each field to fill out, click and complete the corresponding information.
  1. Once the fields are filled out, at the end of the form copy your scanned signature or digital signature. Please adjust the size of the signature on the form.
  2. Finally, you can save them as pdf files and send them through the OJS platform as an attachment.