Modified Regression Approach for Predicting Number of Dengue Fever Incidents in Malang Indonesia

This paper adopted regression approach with Least Square and Natural Logarithmic transformation in response variables to predict the number of Dengue fever attacks in Malang Regency, Indonesia. The prediction involved weather factors. 8 models were prepared, and it was found that the weather factor was the most influential. Some tests, including hypothesis test, were adopted to identify the significance of the model found. The model using response variable with logarithmic natural transformation resulted better model compared to the ones without transformation. It was also supported by the average MAPE of the model that was less than 10%. Therefore, it was identified that the regression approach will work well if both dependent and independent variables have relatively similar variances so that the variability of the dependent variables can be well explained by the independent variable.