Binary Logistic Regression on the Status of Unemployment Rate in West Java Province in 2023

Abstract

Unemployment is a persistent socio-economic challenge that reflects disparities in labour market performance across regions. This study examined the unemployment status of regencies and cities in West Java Province in 2023 using binary logistic regression. The response variable was classified into two categories, namely target achieved and target not achieved, while the explanatory variables consisted of the dependency ratio, senior high school gross enrollment rate, and gross regional domestic product growth rate. Secondary data from 27 regencies/cities were analysed through model estimation, goodness-of-fit assessment, and classification performance evaluation. The estimated model yielded a deviance value of 32.8583 and achieved an overall accuracy of 74.07%, with sensitivity of 30%, specificity of 100%, and an area under the curve of 65%, indicating moderate discriminatory ability. Furthermore, the Press’s Q statistic exceeded the chi-square threshold, suggesting that the classification performance was better than chance. These findings indicate that binary logistic regression provides a useful statistical framework for identifying regional unemployment status and may support evidence-based labour policy evaluation in West Java. However, the relatively low sensitivity suggests the need for additional predictors and further model refinement to improve predictive performance.