Fuzzy linear regression for tuberculosis case notification rate prediction in Surabaya

In this paper we discuss the fuzzy linear regression for predicting the Tuberculosis (TB) case notification rate prediction in Surabaya. The prediction of this disease is very important since TB is among the diseases that carries stigma because it is potentially highly contagious. In this research, we first describe the statistics for the case notification rate (CNR), particularly for the year 2017. We cluster the CNR for man, woman and total with respect to the poverty percentage in each district. Based on this clustering we can map in which district the Tuberculosis CNR and the poverty percentage is high, medium and low. Using the Moran I' statistics, we then tested the spatially dependency among the district. Based on the test result therefore, in this research we also regard the spatial effect of a district to the others in spreading the disease, and model the prediction using fuzzy linear regression. The model can predict good, the mean square error of the model is 835.87 and the mean absolute deviation is 22.46