Applications of Machine Learning Algorithm for Community-Based Health Insurance Classification in Ethiopia
Seyifemickael Amare Yilema1, *, Yegnanew A. Shiferaw2, Yikeber Abebaw Moyehodie1, Setegn Muche Fenta1, Denekew Bitew Belay3, Haile Mekonnen Fenta3
Adopting different machine learning algorithms to the Ethiopian health insurance data.
This study employed the 2019 Ethiopian mini demographic and health survey (EMDHS) data. The community-based health insurance (CBHI) data is used for machine learning (ML) comparisons.
The study's objective is to identify the ML algorithm for predicting health insurance and the important predictors.
Data splitting involves partitioning the data into an explicit training dataset to prepare the model and an unseen test dataset to evaluate the models' performance. ML algorithms such as linear discriminant analysis (LDA), support vector machine with radial basis function (SVM), k-nearest neighbors (KNN), classification and regression trees (CART), and random forest (RF) were employed for predicting health insurance.
The RF model algorithm is the best predicting power with large accuracy of CBHI. Many covariates are employed in the study, but the most important predictors are identified by using RF algorithms. The most important variables are age, wealth, household members and land usage were selected to be the top four important variables for predicting health insurance.
This study will be helpful for health insurance professionals and policy makers to develop a system for adopting necessary interventions. Further research will be conducted for ML-based classifiers and deep learning to predict health insurance.
* Address correspondence to this author at the Department of Statistics, College of Natural and Computational Science, Debre Tabor University, Debre Tabor, P.O.Box: 272, Ethiopia; E-mail: firstname.lastname@example.org