Forecasting the tuberculosis morbidity rate in Indonesia using temporal convolutional neural network and exponential smoothing

Abstract

Tuberculosis (TB) morbidity rate in Indonesia shows the number of population in Indonesia who suffer from TB. The TB morbidity rate can be used by insurance companies to predict a person’s risk of TB so that insurance companies can determine the premiums that will be charged to insurance applicants based on the risks. Thus, the ability to estimate the TB morbidity rate accurately is essential for insurance companies to be able to determine the right premium amount while remaining competitive. This study compared two models that can be used to predict TB morbidity rate in Indonesia. The model was built using the temporal convolutional neural network (TCNN) and exponential smoothing methods. The data analyzed in this study are data obtained from the official website of the Ministry of Health of the Republic of Indonesia. Before the model was built, the data used in this study were compiled into training and validation datasets. The model is built using a training dataset and validated using the validation dataset. The results of the model’s validation are then evaluated and compared based on the value of the mean squared error (MSE). The result of this study shows that the TCNN model provides lower MSE compared to exponential smoothing.