@article{paperid:1030915, author = {Fezeh Zahedi Fard and Salehi, Mahdi}, title = {A Comparison of Data Mining Techniques for Going Concern Prediction}, journal = {International Journal of Research in Commerce, IT & Management}, year = {2012}, volume = {2}, number = {4}, month = {November}, issn = {2231-5756}, pages = {14--19}, numpages = {5}, keywords = {Going concern is one of the fundamental concepts concerning auditing and accounting. Since financial statements contain a potentially large volume of diversified information; sometimes firm’s going concern status evaluation is a complex and critical process and the complexity of this issue has led to development of numerous models for going concern prediction (GCP). In this paper we proposed a novel approach using Imperialistic Competition Algorithm (ICA). In addition; we have presented more advanced data mining techniques; like Adaptive Network Based Fuzzy Inference Systems (ANFIS) and Support Vector Data Description (SVDD) for GCP. For this purpose; after data collection we have selected the final variables from among of 42 variables based on feature selection method using stepwise discriminant analysis (SDA). In the second stage we have applied 10-fold cross-validation to find out the optimal model. Results of three models statistically have been compared by McNemar test. Our empirical experiment indicates that ICA is more efficient than ANFIS and SVDD; but ANFIS does not significantly differ from SVDD. The ICA model reached 99.85 and 99.33 percent accuracy rates so as to training and hold-out data.}, }