Stunting Classification in Children's Measurement Data Using Machine Learning Models

Abstract

The study conducted a stunting classification of measurement data for children under 5 years old. The dataset has attributes such as: gender, age, weight (BB), height (TB), weight / height (BBTB), weight / age (BBU), and height / age (TBU). The research uses the CRISP-DM methodology in processing the data. The data were tested on several classification models, namely: logistic regression (LR), linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), k-nearest neighbor (KNN), classification and regression trees (CART), nave bayes (NB), support vector machine - linear kernel (SVM-Linear), support vector machine - rbf kernel (SVM-RBF), random forest classifier (RPC), adaboost (ADA), and neural network (MLPC). These models were tested on the dataset to find out the best model in accuracy. The test results show that SVM-RBF produces an accuracy of 78%. SVM-RBF has consistently been at the highest accuracy in several tests. Testing through k-fold cross validation with k=10.