A Comparative Study of Machine Learning Algorithms in Classifying and Predicting Exclusive Breastfeeding Infants in Bangladesh

This study addresses the analysis of exclusive breastfeeding (EBF) with the help of infants’, mothers', and household characteristics associated with EBF. Data were collected from the Bangladesh Demographic and Health Survey (BDHS), in 2017-18 and 2022. The objective of this study is to investigate different machine learning algorithms, for instance, Artificial Neural Networks (ANN), K-nearest Neighbors (KNN), Random Forests (RF), Classification and Regression Trees (CART), Support Vector Machines (SVM), Linear Discriminant Analysis (LDA), Logistic Regression (LR), Naive Bayes (NB), XGBoost (XGB) and AdaBoost (ADA) to classify and predict exclusive breastfeeding status of an infant. The classification accuracy, sensitivity, specificity, kappa statistic, and area under the receiver operating characteristic curve (AUROC) were used to evaluate and compare the performance of these mentioned models. Among the ten machine learning models evaluated, the best-performing models were Random Forest (accuracy: 91.2%, sensitivity: 92.0%, specificity: 90.7%, AUROC: 0.966) and Extreme Gradient Boosting (accuracy: 89.4%, sensitivity: 93.7%, specificity: 86.7%, AUROC: 0.947). The worst-performing models were Naive Bayes (accuracy: 69.7%, sensitivity: 44.0%, specificity: 87.0%, AUROC: 0.729) and Logistic Regression (accuracy: 71.9%, sensitivity: 57.5%, specificity: 81.6%, AUROC: 0.772). Overall, Random Forest and Extreme Gradient Boosting emerged as robust models for predicting exclusive breastfeeding.