Predicting the Filipino Household Income Using Naive Bayes Classification Algorithm

Abstract

The Family Income and Expenditure Survey (FIES) in the Philippines provides crucial data on household income and expenses. This study utilizes the Naive Bayes algorithm to predict Filipino economic class using household expenditure and income variables. The research aims to contribute to poverty reduction efforts by providing a predictive model for identifying vulnerable households and designing appropriate interventions. Data preprocessing steps, including data cleaning, transformation, and analysis, were performed before feature selection and modelling. Predictive models using Naive Bayes were evaluated and validated, with accuracy measured using a confusion matrix. Results show high accuracy rates, with bagging (93%) and boosting (89%) ensemble techniques used for model implementation. Findings can potentially support local government units in poverty reduction programs and policymaking. Future research could explore other machine learning algorithms and consider additional parameters to further improve model accuracy using the increasing data from the FIES datasets provided by the Philippine Statistics Authority.