Forecasting International Migration Based on a Hybrid Fuzzy and Bayesian Method
Tarih ve Saat
18 Şubat 2019 - 16:00
YerMühendislik Binası, A 511
Forecasting international migration is one of the vital elements in demographic analysis since it plays a significant role in shaping the socio-economic structures of countries. The current interest in forecasting migration has led to several migration theories as well as deterministic and stochastic forecasting methods. These methods are based on strict subjective or statistical assumptions which may not always be met. Moreover, the exact values of migrants are seldom known due to data recording and collection errors; thus, there is a significant amount of vagueness and uncertainty in migration data. In this study, to deal with the uncertainties in migration values, a hybrid method integrating fuzzy set theory and Bayesian forecasting is proposed for forecasting age-specific migration. The proposed method models the observed migration values via fuzzy regression and an unconstrained nonlinear optimization model, and forecasts the future fuzzy migration values using Bayesian time series models. The proposed method is applied on emigration and immigration data of Finland, in which annual age-specific migration values for 2011-2025 are forecasted using the fuzzy estimates for 1990-2010. The results are compared with an existing Bayesian migration forecasting method outputs and the numerical findings display that the proposed hybrid method is superior to the existing one in forecasting age-specific migration values within significantly narrower prediction intervals.
Keywords: Fuzzy Modeling, Bayesian Forecasting, Unconstrained Nonlinear Optimization, Fuzzy Regression, Migration Analysis