Title: Application of machine learning algorithms for population forecasting

Authors: Fatih Veli Şahinarslan; Ahmet Tezcan Tekin; Ferhan Çebi

Addresses: Department of Management Engineering, Istanbul Technical University, Istanbul, 34367, Turkey ' Department of Management Engineering, Istanbul Technical University, Istanbul, 34367, Turkey ' Department of Management Engineering, Istanbul Technical University, Istanbul, 34367, Turkey

Abstract: In this study, different machine learning algorithms were used to forecast population; extreme gradient boosting, CatBoost, linear regression, ridge regression, Holt-Winters, exponential, autoregressive integrated moving average (ARIMA) and prophet prediction model. Models were trained using 1595 different demographic indicators of 262 different countries between 1960 and 2017. When the performance of algorithms was compared, the extreme gradient boosting model was the most successful among all models. Besides, the total population of Turkey in 2017 estimated by pre-trained machine learning algorithms were compared with the result predicted by Cohort component method. Results showed that machine learning algorithms performed better than the demographic model.

Keywords: population forecasting; machine learning; ensemble learning; cohort component method; regression; time-series models.

DOI: 10.1504/IJDS.2021.122770

International Journal of Data Science, 2021 Vol.6 No.4, pp.257 - 270

Received: 04 Jul 2020
Accepted: 20 Oct 2021
Published online: 10 May 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article