Institute media
The Department of Financial Studies the Post-graduate Institute for Accounting and Financial Studies at the University of Baghdad discussed the research titled (Forecasting Tax Revenues Using DATA Mining Techniques/An Applied research at the General Commission for Taxes ) for the student Abe Dher Nameer Qasim to obtain a higher diploma equivalent to a master’s degree in taxes and grants its holder All rights and privileges of a master’s degree.
This study aimed to select the best tax revenue forecasting model from among five established data mining methods: Artificial Neural Networks (ANNs), Convolutional Neural Networks (CNNs), Long-Short-Term Memory Networks (LSTMs), Decision Tree Regression (DTR), and Support Vector Regression (SVR). The goal was to improve the accuracy of future tax revenue forecasts, as the research problem stemmed from the inaccuracy of tax revenue forecasting at the General Authority for Taxes due to reliance on subjective estimation rather than modern scientific methods.
The study concluded that the Long-Short-Term Memory Network (LSTM) model was superior in providing more efficient and accurate future tax revenue forecasts for the years 2026-2030 compared to the other data mining methods. It achieved the lowest error metrics and the highest correlation, highlighting the model’s contribution to supporting financial planning and the formulation of tax policies based on sound scientific principles.
