OPTIMALISASI REGRESI BINOMIAL NEGATIF DENGAN REGULARISASI LASSO MENGGUNAKAN K-FOLD CROSS VALIDATION (STUDI KASUS : JUMLAH KASUS DBD DI JAWA TENGAH TAHUN 2024)NIM.: 22106010081Robit Ma’rufi SabilaauthorRegression is one of the statistical methods used to analyze the relationship between dependent and independent variables. For count data, the Poisson regression model is commonly used. Poisson regression assumes equidispersion, where the mean and variance of the dependent variable are equal. However, in practice, Poisson regression often encounters overdispersion, where the variance is greater than the mean, making Poisson regression less appropriate. One method for addressing overdispersion is Negative Binomial regression, which has a dispersion parameter that can accommodate overdispersion in the data. In addition to overdispersion, multicollinearity, characterized by high correlations among independent variables, can cause regression coefficients to become unstable. The use of high-dimensional data can also increase the risk of overfitting. Therefore, a variable selection method is needed to produce a simpler and more efficient model. To address these issues, Lasso regularization is applied by imposing an L1 penalty on the regression coefficients, thereby producing more stable parameter estimates while simultaneously performing variable selection. This study uses the number of Dengue Hemorrhagic Fever (DHF) cases in Central Java in 2024 as the dependent variable and 12 independent variables related to health, sanitation, and socioeconomic factors. This study aims to apply Negative Binomial regression with Lasso regularization using K-Fold Cross Validation. The results show that the optimal λ value of 0.4247 selects one variable, namely the Number of Community Health Centers (JSKP), resulting in a more parsimonious and efficient model.515.6 Metode Analitik - Matematika2026-08-03UIN SUNAN KALIJAGA YOGYAKARTA;FAKULTAS SAINS DAN TEKNOLOGIThesis