Modelling and Forecasting Foreign Direct Investment - A Comparative Application of Machine Learning Based Evolutionary Algorithms Hybrid Models

Aug 4, 2026

The study evaluates whether genetic algorithms (GA) can enhance forecasting accuracy of nonlinear supervised machine learning models for foreign direct investment (FDI). Evolutionary-algorithm hybrid models are built by combining GA with artificial neural networks (ANN-GA) and support vector regression (SVR-GA), and are compared with benchmark traditional models: ANN, SVR, and least squares support vector regression (LSSVR). Monthly FDI time series data with macroeconomic drivers (GDP, inflation rate, exchange rate) span January 1970 to June 2019 (594 observations). Nonlinearity tests confirm nonlinear characteristics, and error comparisons supported by Diebold-Mariano results show ANN as best, while GA-based hybrids do not significantly improve performance.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages159

Reading Time6 h 41 min

AccessFree to download

Declaration

Acknowledgements

Dedication

Preface

Abstract

Research purpose and method

Data and nonlinearity testing

Model comparison and results

Conclusions and recommendations

Modelling and Forecasting Foreign Direct Investment: A Comparative Application of Machine Learning Based Evolutionary Algorithms Hybrid Models by Mr. Mogari Ishmael Rapoo [orcid.org/0000-0003-0771-7461](orcid.org/0000-0003-0771-7461) Thesis submitted for the degree Doctor in Business Statistics in Economic and Management Sciences on Mafikeng Campus at the North-West University Promoter: Prof Elias Munapo Co-promoter: Prof Martin M. Chanza Examination: November 2022 Student number: 23809213 DECLARATION I declare that this thesis is the result of my own investigation unless stated otherwise. I also declare that it has never been submitted previously as a whole or in part for any other degree whatsoever to the North-West University or any institution. Mogari Ishmael Rapoo Rapoo MI 06/12/2022 (23809213) Signature Date ACKNOWLEDGEMENTS I would want to thank the Almighty God who enabled me and taken me under his mighty wings to understand, undertake and complete this work through everything else as part of the academic requirements for my PhD programme. His love and mercy have showered me throughout this difficult yet fulfilling journey. My sincere gratitude goes to both my promoters; Prof E Munapo and Dr MM Chanza, for providing such mind stimulating guidance and for supporting me throughout this thesis. Without their commitment in making sure that this thesis is of good quality , I would never have been able to finish this thesis. Moreover, to Mr Monchwe , thank you for dedicating your time with your coding skills. Secondly, I would like to extend my gratitude to my family, especially my father, my mother, my girlfriend and son, brothers and uncles, for their support throughout this thesis. DEDICATION To my entire family, those who departed this world and those who are still alive , the love and support you show me is, indeed, second to none. PREFACE The purpose of this thesis is to determine whether a genetic algorithm (GA) can improve the forecasting performance of nonlinear supervised machine learning models in modelling and forecasting foreign direct investment. Hybrid models of genetic algorithm-based artificial neural network and genetic algorithm-based support vector regression employed to improve the forecasting performance and accuracy of the models. The past three years have been a challenging time with both ups and downs, a time of great benefit for me even through the hardest moments. Fortunately, I was never alone in this whole time, and I was accompanied by expertise in my promoters who were always coaching me and helping me and for that, I am truly thankful. To my entire family, the support and love you all showed me were deeply noticed. To my promoters, your knowledge, guidance and persistent pushing have enabled me to finish this thesis and I am truly thankful. Many thanks to everyone, Colette my girlfriend, Tshiamo my son, Mr Maleke my father, Ms Rapoo my mother, Mr Rapoo my brother and Mr Legae my uncle for all the encouragement and support. This thesis will contribute to the academic discipline and to those in the field of research in machine learning, evolutionary algorithms and for time series forecasting. It should also be of interest to scholars of machine learning and to those who want to contribute to the field of machine learning. ABSTRACT The study aims to determine whether genetic algorithms can improve the forecasting accuracy of machine learning models in modelling and forecasting foreign direct investment (FDI) . The use of evolutionary algorithms for FDI modelling and forecasting is relatively unexplored. The research therefore employs benchmark models (ANN, SVR, and LSSVR) to assess the forecasting performance of hybrid models (ANN-GA and SVR-GA) in modelling and forecasting FDI time series data. Monthly time series data of FDI (dependent variable) and GDP , inflation rate, and exchange rate (explanatory variables) were collected from Worl

FAQs

What is the main purpose of the thesis?

To determine whether genetic algorithms can improve forecasting accuracy of nonlinear supervised machine learning models for modelling and forecasting foreign direct investment (FDI).

Which models are used to compare forecasting performance?

Benchmark models include ANN, SVR, and LSSVR, while hybrid models use GA with ANN (ANN-GA) and GA with SVR (SVR-GA).

What do the results and statistical tests conclude?

ANN achieved the lowest error measures and best forecasting accuracy, while GA-based hybrid models did not significantly outperform traditional models; Diebold Mariano test supports these ordering results.

Aug 4, 2026

Upload by Evangeline