Abstract:
The Poverty Gap Index (PGI) is the average expenditure gap of each poor population
towards the poverty line. This study aims to model PGI data using binary logistic
regression with a classical approach using the Maximum Likelihood Estimation (MLE)
method and a Bayesian approach using the Markov Chain Monte Carlo (MCMC)
method. MCMC is a popular method for obtaining information about the distribution,
especially for estimating the posterior distribution in Bayesian inference with the
Metropolis-Hasting algorithm. Factors that have a significant influence on the IKK in
Sumatera using the Bayesian approach and the classical approach are the same, namely
Life Expectancy and per capita expenditure. Based on the results of the classification
with training data of 80% and test data of 20% a classification accuracy of 62,50%.
