Development of a Prediction Model on Demographic Indicators based on Machine Learning Methods: Azerbaijan Example

The accuracy of population forecasts is one of the most important calculations in demography statistics. However, traditional demographic methods used in population projections are tend to produce biased results. The need for accurate prediction of future behavior in a number of areas require the application of reliable and efficient methods. Recently, machine learning (ML) models have emerged as a serious competitor to classical statistical models in the forecasting community. In this study, the performance and capacity of the four different ML models such as Random forest (RF), Decision tree (DT), Linear regression (LR) and K-nearest neighbors (KNN) to the prediction of population has been examined. The aim of the study is to find the best performing regression model among these machine learning algorithms for forecasting of population. The data were collected from the State Statistical Committee of the Republic of Azerbaijan website were used for the analysis. We used five metrics such as mean absolute percentage error (MAPE), mean absolute error (MAE), root mean squared error (RMSE), mean square error (MSE) and R-squared to compare the predictive ability of the models. As the result of the analysis, it has been known that the all ML models showed high results with correlation coefficient of 0.985 - 0.996. Also the KNN and RF prediction models showed the lowest root mean square deviation, means square error and mean absolute error values compared to other models. By effectively using the advantage of the ML algorithms, the forecast of population growth the near future can be observed objectively, and it can provide an objective reference to the strategic planning in the public and private sectors, particularly in education, health and social areas.