Prediction of Infant Size at Birth Using Principal Components based Artificial Neural Network
Click here to download PDF

Keywords

Factor analysis
Artificial Neural Network
Prediction
Infant
Multicollinearity

How to Cite

Usman, U., Suleiman, S., Muhammad, A. B., & Ibrahim, M. B. (2023). Prediction of Infant Size at Birth Using Principal Components based Artificial Neural Network. ARID ZONE JOURNAL OF ENGINEERING, TECHNOLOGY AND ENVIRONMENT, 19(2), 331-340. Retrieved from https://azojete.com.ng/index.php/azojete/article/view/742

Abstract

Birth size plays a major role in child’s mental development, future physical growth, and survival. The average length of full-term babies at birth is 20 in. (51 cm), although the normal range is 46 cm (18 in.) to 56 cm (22 in.). In the first month, babies typically grow 4 cm (1.5 in.) to 5 cm (2 in.). Principal Component Analysis (PCA) is a dimension-reduction tool that can be used to reduce a large set of variables to a small set that still contains most of the information in the large set. In neural networks multicollinearity may increase difficulties in data processing; too much feature inputs will also cause the time-consuming training process, prevent the constringency of the training work, and may eventually affect the network performance. This research investigates the effect of some parents’ socioeconomic status on the sizes of infants at birth in Zamfara State. Several factors have been identified to have effect on the sizes of infant at birth which is part of major public health challenges in developing countries like Nigeria.  Previous studies considered only some maternal factors responsible for the infant sizes at birth. This current study explored the parents’ socioeconomic Status and environmental factors that contributes to the status of the infant sizes at birth using factor and neural network analysis.  The analysis was carried out using data collected from Federal Medical Centre, Gusau and Ahmad Sani Yariman Bakura Specialist Hospital Gusau, Zamfara State to explore the association between the afore mentioned factors. Multicollinearity results indicated that there is significant correlations among independent variables, factor analysis through principal extraction methods reduced the original 14 correlated variables in to five uncorrelated factors. However, artificial neural network performs better in the presence of multicollinearity as it produces lower mean square error (MSE). It also had a better classification accuracy of 83.5% as compared to 65.9% for Gender of the baby.

 

Click here to download PDF
Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Copyright (c) 2023 Arid Zone Journal of Engineering, Technology and Environment published by University of Maiduguri, Nigeria