OBJECTIVE: The aim of this study was is to predict malnutrition status in under-five children in Bangladesh by using various machine learning (ML) algorithms. METHODS: For analysis purposes, the nationally representative secondary records from the 2014 Bangladesh Demographic and Health Survey (BDHS) were used. Five well-known ML algorithms such as linear discriminant analysis (LDA), k-nearest neighbors (k-NN), support vector machines (SVM), random forest (RF), and logistic regression (LR) have been considered to accurately predict malnutrition status among children. Additionally, a systematic assessment of the algorithms was performed by using accuracy, sensitivity, specificity, and Cohen's κ statistic. RESULTS: Based on various performance parameters, the best results were accomplished with the RF algorithm, which demonstrated an accuracy of 68.51%, a sensitivity of 94.66%, and a specificity of 69.76%. Additionally, a most extreme discriminative ability appeared by RF classification (Cohen's κ = 0.2434). CONCLUSION: On the basis of the findings, we can presume that the RF algorithm was moderately superior to any other ML algorithms used in this study to predict malnutrition status among under-five children in Bangladesh. Finally, the present research recommends applying RF classification with RF feature selection when the prediction of malnutrition is the core interest.
OBJECTIVE: The aim of this study was is to predict malnutrition status in under-five children in Bangladesh by using various machine learning (ML) algorithms. METHODS: For analysis purposes, the nationally representative secondary records from the 2014 Bangladesh Demographic and Health Survey (BDHS) were used. Five well-known ML algorithms such as linear discriminant analysis (LDA), k-nearest neighbors (k-NN), support vector machines (SVM), random forest (RF), and logistic regression (LR) have been considered to accurately predict malnutrition status among children. Additionally, a systematic assessment of the algorithms was performed by using accuracy, sensitivity, specificity, and Cohen's κ statistic. RESULTS: Based on various performance parameters, the best results were accomplished with the RF algorithm, which demonstrated an accuracy of 68.51%, a sensitivity of 94.66%, and a specificity of 69.76%. Additionally, a most extreme discriminative ability appeared by RF classification (Cohen's κ = 0.2434). CONCLUSION: On the basis of the findings, we can presume that the RF algorithm was moderately superior to any other ML algorithms used in this study to predict malnutrition status among under-five children in Bangladesh. Finally, the present research recommends applying RF classification with RF feature selection when the prediction of malnutrition is the core interest.
Authors: S M Jubaidur Rahman; N A M Faisal Ahmed; Md Menhazul Abedin; Benojir Ahammed; Mohammad Ali; Md Jahanur Rahman; Md Maniruzzaman Journal: PLoS One Date: 2021-06-17 Impact factor: 3.240