Machine Learning Approaches in Classifying Income Levels

Emely L. Laspiñas; John Vianne B. Murcia

Autori

Emely L. Laspiñas College of Development Management, University of Southeastern Philippines – Mintal Campus, Mintal, Davao City 8000, Philippines
John Vianne B. Murcia [1]- College of Development Management, University of Southeastern Philippines – Mintal Campus, Mintal, Davao City 8000, Philippines | [2]- College of Business Administration Education, University of Mindanao, Bolton Street, Davao City 8000, Philippines https://orcid.org/0000-0002-8839-204X

Cuvinte cheie:

Machine learning, Weka, Income levels, Classifiers, Classification

Rezumat

This study compares the predictive accuracy of six machine learning classifiers – Logistic, Decision Tree (J48), RandomForest, Random Tree, IBk (k-NN), and NaiveBayes – for estimating adult income. Utilizing metrics such as true positive (TP) rate, false positive (FP) rate, precision, recall, and the F-measure, the performance of these classifiers was evaluated. Based on the results, RandomForest and Random Tree classifiers demonstrated the highest efficacy across all metrics. Nonetheless, other classifiers, such as Decision Tree and IBk, demonstrated promise, especially when the parameters were modified. The findings highlight the importance of model selection and fine-tuning in predictive modeling. These findings have significant ramifications for income forecasting, highlighting the capacity of machine learning to facilitate accurate socioeconomic forecasting. The study's results provide vital guidance for deploying the most appropriate classifier based on the specifics of the income dataset and the prediction task.