Publisher

ITSITS

Journal IPTEK The Journal for Technology and ScienceIPTEK The Journal for Technology and Science Abstract

Taxpayer non-compliant behavior impacts Motor Vehicle Tax (MVT) revenues, potentially hindering regional development targets. This research aims to analyze taxpayer behavior in predicting future MVT payments – whether they are compliant, late, or non-payment. The proposed approach involves analyzing taxpayer behavioral features and utilizing an ensemble classifier based on Weighted Majority Voting (WMV). WMV was developed using GridSearchCV to optimize hyperparameters and improve model accuracy. Feature ablation analysis was conducted to understand the contribution of each feature to model performance. The research results demonstrate that the WMV method outperforms individual classifiers, achieving an accuracy of 96.247% in predicting MVT payments based on taxpayer behavior.

Conclusion

This research successfully analyzed taxpayer behavior to predict MVT payments using a Weighted Majority Voting Ensemble method.The proposed approach, optimized with GridSearchCV, demonstrated superior accuracy compared to individual classifiers, reaching 96.Feature ablation analysis provided valuable insights into the contribution of each feature to the prediction process.Overall, the study offers a promising strategy for improving MVT revenue collection and supporting regional development goals.

Future Research

Future research should explore expanding the dataset with additional taxpayer behavioral data and incorporating diverse machine learning methods to enhance prediction accuracy. Investigating the impact of socio-economic factors on MVT payment behavior could provide a more comprehensive understanding of taxpayer compliance. Furthermore, developing a real-time prediction system integrated with regional tax administration systems could enable proactive interventions to encourage timely payments and minimize revenue loss. These advancements will contribute to more effective tax collection strategies and improved regional financial planning, building upon the findings of this study and addressing potential limitations by incorporating a broader range of influencing variables and practical applications.

References

1. Sigmis: A Feature Selection Algorithm Using Correlation Based Method - E. Chandra Blessie, E. Karthikeyan,... journals.sagepub.com/doi/10.1260/1748-3018.6.3.385Sigmis A Feature Selection Algorithm Using Correlation Based Method E Chandra Blessie E Karthikeyan journals sagepub doi 10 1260 1748 3018 6 3 385
2. AutoAblation | Proceedings of the 1st Workshop on Machine Learning and Systems. autoablation proceedings... dl.acm.org/doi/10.1145/3437984.3458834AutoAblation Proceedings of the 1st Workshop on Machine Learning and Systems autoablation proceedings dl acm doi 10 1145 3437984 3458834

Metadata

File size	289.11 KB
Pages	13
DMCA	Report