DATA ANALYSIS OF DENGUE FEVER CASE STUDY ON ONE OF THE HOSPITALS IN BANDUNG CITY

• Main Author: Luthfi Ibrahim, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/79623
• Institution: Institut Teknologi Bandung

This research seeks to analyze dengue fever case data with the aim to describe the characteristics, patterns, and wawasans of dengue fever case data obtained from one of the hospitals in Bandung City. This data-centered research has three different approaches, namely correlation analysis between variables in the data, spatial analysis of the distribution of cases in Bandung City, and exploration of a suitable model to serve as a dengue fever case prediction model for this data. The correlation analysis was divided into three: numerical pairs, categorical pairs, and categorical and numerical pairs. Pearson, Spearman, and Kendall's Tau correlations were used for numerical pairs, while chi-square and Cramer's V independence tests were applied to categorical pairs. Levene's test, normality test, and data type identification were used on pairs consisting of categorical and numerical variables. It was found that the variables in the data had very little correlation, except for the location variable. Furthermore, spatial analysis was conducted to identify klasters of spread and high risk areas. Moran's auto correlation I and Getis Ord G* were used in this approach to understand the spatial pattern of dengue fever cases. The results of the spatial analysis showed a trivial result, namely that Coblong sub-district became the center of dengue fever case distribution. Finally, prediction model exploration was conducted using four models: multilinear regression, negative binomial model, random forest, and XGBoost Regressor using MSE and MAPE evaluation parameters. The goal of these models is to produce accurate predictions related to dengue fever cases. It is found that XGBoost Regressor is the best model to predict dengue fever cases in this data.