Author Login IJIRT Partner Network (IPN) Login

Advancing Income Tax Fraud Detection: A Comprehensive Review Of Machine Learning And Deep Learning Models

  • Unique Paper ID: 170876
  • Volume: 11
  • Issue: 7
  • PageNo: 923-933
  • Abstract:
  • As digital financial systems continue to expand and transaction volumes rise, the risk of fraud has increased considerably. This paper explores how Machine Learning (ML) and Deep Learning (DL) techniques can be utilized to detect income tax fraud, emphasizing important metrics like accuracy, precision, recall, and F1-score to assess the performance of the models. Core Methodologies: The model applied is those of the random forest, XGBoost and Decision Tree Classifier, together with CNN and LSTM architectures in deep learning models. In addition to these improvements in models, preprocessing is done that involves feature engineering, using one-hot encoding of categorical features, and then normalization. The study focuses on a highly imbalanced synthetic dataset with a fraud rate of 0.9% and 1 million entries, tackling issues related to computational complexity and class imbalance with customized strategies. Performance Insights: Random Forest and XGBoost proved to be better since the F1-scores obtained were 0.9472 and 0.9522, respectively. However, for the same experiment, the performance of Decision Tree classifiers was competitive with an F1-score of 0.9407 and relatively very less computation time, and for CNNs and LSTMs, it could not be impressive as there is less recall and relatively higher computation, and hence F1-scores 0.3341 and 0.5110 respectively. This paper discusses the strengths and weaknesses of each model in fraud detection, discussing the implications of imbalanced datasets and computational trade-offs. It provides actionable insights for scholars, policymakers, and industry stakeholders who are looking to improve financial stability through advanced AI-driven fraud detection systems. Future recommendations include optimizing DL architectures for recall and leveraging ensemble approaches to further improve detection accuracy and efficiency.
Download article email to a friend

Cite This Article

  • ISSN: 2349-6002
  • Volume: 11
  • Issue: 7
  • PageNo: 923-933

Advancing Income Tax Fraud Detection: A Comprehensive Review Of Machine Learning And Deep Learning Models

Related Articles

Impact Factor
8.01 (Year 2024)
UGC Approved
Journal no 47859

Join Our IPN

IJIRT Partner Network

Submit your research paper and those of your network (friends, colleagues, or peers) through your IPN account, and receive 800 INR for each paper that gets published.

Join Now

Latest Publication

Archive

View More

Recent Conferences

NCSEM 2024

National Conference on Sustainable Engineering and Management - 2024 Last Date: 15th March 2024

Submit inquiry