A Hybrid Approach of Stepwise Regression, Logistic Regression, Support Vector Machine, and Decision Tree for Forecasting Fraudulent Financial Statements

Table 5

SVM cross-validation results.

SVM model			Predict value		Hit ratio	Type I error	Type II error
			Non-FFS	FFS
Actual value	CV1	Non-FFS	26	2	73.21%	7.14%	46.42%
		FFS	13	15
	CV2	Non-FFS	26	2	71.43%	7.14%	50.00%
		FFS	14	14
	CV3	Non-FFS	26	2	71.43%	7.14%	50.00%
		FFS	14	14
	Average	Non-FFS	26	2	72.02%	7.14%	48.81%
		FFS	14	14