Enhanced recession forecasting using artificial neural network algorithm
Date
2018-09
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Federal University of Technology, Owerri
Abstract
Recession is a global concern, as nations, industries and individuals are affected. Application of machine learning in finance is not novel, it strength in this field is shown in this thesis. Several literature on the use of artificial neural network and key financial indicators were “x-rayed” with the peculiar nature of the Nigerian state at the background. From the thirty one variables initially taken from the Central bank of Nigeria statistics portal and the Nigerian Bureau of Statistics and subjected to a genetic algorithm, twelve most correlated variables to economic growth from four sectors were used. These were used to train and test the Neural Network algorithm and then compared with known regression and generalized linear model results. Findings from this study revealed that Artificial Neural Network outperformed the other models in terms of speed of prediction, robustness of algorithm and accuracy. The algorithm developed is able to make monthly recession forecasts after being trained and tested with Data from 2010 to 2017. The performance of the neural network far outweighed the Regression and the Generalized Linear model as its Mean Square Error was approximately 4 while the generalized neural network was 25 and was 14 for the regression model.
Description
This thesis is for the award of Master of Science (MSc.) in Information Management Technology
Keywords
Artificial neural network, model, recession probability, performance indicators, machine learning, Department of Information Management Technology
Citation
Ndubuisi, J. E. (2018). Enhanced recession forecasting using artificial neural network algorithm [Unpublished Master's Thesis]. Federal University of Technology, Owerri, Nigeria