<![CDATA[Tax evasion by Corporate Taxpayers is a significant challenge for state revenue. This study aims to develop a tax evasion behavior prediction system using a deep learning approach based on Convolutional Neural Network (CNN). The objects of this study were 50 Corporate Taxpayers registered in Lamongan Regency during the period 2021 to 2024. The data used are secondary data from the manufacturing, trade and service sectors which include financial indicators such as Effective Tax Rate (ETR), total assets and reporting compliance. The data is then labeled into two categories, namely "risky" and "not risky" to train the CNN model. The test results show that the developed model is able to classify tax evasion behavior with an average accuracy of 84.29%. The indicators "Total Assets and Liabilities" and "Reporting Compliance" are the predictors with the highest accuracy, respectively at 86.00% and 85.71%. This study concludes that the CNN model is proven to be effective and relevant for use as a decision support system for tax authorities in detecting potential tax avoidance earlier and more efficiently.]]>
Copyrights © 2025
