Estimation of Tax Loss and Evasion in Turkey with Data Mining Process

Authors : Derya Şencan

Pages : 11-24

Doi:10.29157/etusbed.1292006

View : 256 | Download : 149

Publication Date : 2023-10-31

Article Type : Research Paper

Abstract :Tax losses and evasion are one of the biggest problems in our country as well as all over the world. In addition to audits, statistical techniques and machine learning algorithms are also of great importance in detecting tax losses and evasion. In this study, the tax loss and evasion rate has been estimated by the data mining process depending on factors such as inflation rate, unemployment, tax burden, trade openness, economic growth (GDP), and the size of government. Twelve modeling techniques were used in the data mining process. The results obtained from each model were compared and the best model was determined using some statistical indicators. Accordingly, the Gaussian processes model gave the most successful result in estimating tax loss and evasion rate, with R2, MAE and RMSE values of 0.931, 0.2356 and 0.2473, respectively. The weight values of the variables affecting the tax loss and evasion rate were determined by sensitivity analysis. It has been observed that the factors with the highest positive effect on tax losses and evasion are unemployment and inflation rates. These factors are followed by tax burden and GDP values. It was seen that the size of government and the trade openness factors had a negative effect on the tax loss and evasion rate. It is thought that the results obtained from the study will contribute to the estimation of the tax loss and evasion rate in our country.
Keywords : Vergi kaybı, Vergi kaçağı, Veri madenciliği

ORIGINAL ARTICLE URL