Using Deep Learning Algorithms for CPAs’ Going Concern Prediction

Certified public accounts’ (CPAs) audit opinions of going concern are the important basis for evaluating whether enterprises can achieve normal operations and sustainable development. This study aims to construct prediction models help CPAs auditors make more effective/correct judgments on opinion decisions by deep learning algorithms, using following methods: neural networks (DNN), recurrent network (RNN), classification regression tree (CART). The samples this companies listed Taiwan Stock Exchange Taipei Exchange, a total 352 companies, including 88 with doubt 264 (with no doubt). data from 2002 2019 taken Economic Journal (TEJ) Database. According empirical results, variables selected CART modeling RNN, CART-RNN model has highest accuracy (the test dataset is 95.28%, average 93.92%).