Governance of artificial intelligence applications in a business audit via a fusion fuzzy multiple rule-based decision-making model

Hu, Kuang-Hua; Chen, Fu-Hsiang; Hsu, Ming-Fu; Tzeng, Gwo-Hshiung

doi:10.1186/s40854-022-00436-4

Financial Innovation

Table 4 The compared results of four approaches (The clustering approach is FKM)

From: Governance of artificial intelligence applications in a business audit via a fusion fuzzy multiple rule-based decision-making model

Condition (Summation of Cov. and Acc.) (Rank)	Friedman test (P value)
K = 2 (We assume this problem is a two-class label classification task)
Multiplication strategy: Increasing 50% of instances in each class label
DRSA (1.75) (1) > CART(1.67) (2) > RSA(1.61) (3) > DT(1.55) (4)	0.000***
Multiplication strategy: Increasing 100% of instances in each class label
DRSA (1.78) (1) > CART (1.72) (2) = RSA (1.71) (2) > DT(1.64) (4)	0.042**
Multiplication strategy: Increasing 150% of instances in each class label
DRSA (1.81) (1) > RSA (1.75) (2) > CART (1.70) (3) > DT(1.67) (4)	0.041**
Multiplication strategy: Increasing 200% of instances in each class label
DRSA(1.83) (1) > RSA (1.77) (2) > CART(1.73) (3) > DT(1.68) (4)	0.031**
K = 3 (We assume this problem is a three-class label classification task)
Multiplication strategy: Increasing 50% of instances in each class label
DRSA(1.77) (1) > RSA (1.73) (2) > CART (1.69) (3) > DT(1.66) (4)	0.031**
Multiplication strategy: Increasing 100% of instances in each class label
DRSA(1.79) (1) = RSA (1.78) (1) > CART (1.72) (3) > DT(1.69) (4)	0.000***
Multiplication strategy: Increasing 150% of instances in each class label
DRSA (1.81) (1) > RSA (1.78) (2) > CART(1.74) (3) > DT(1.71) (4)	0.000***
Multiplication strategy: Increasing 200% of instances in each class label
DRSA (1.84) (1) > RSA (1.79) (2) > CART (1.76) (3) > DT(1.73) (4)	0.000***
K = 4 (We assume this problem is a four-class label classification task)
Multiplication strategy: Increasing 50% of instances in each class label
DRSA (1.63) (1) > RSA (1.58) (2) > CART (1.55) (3) = DT(1.54) (3)	0.042**
Multiplication strategy: Increasing 100% of instances in each class label
DRSA (1.67) (1) = RSA (1.66) (1) > CART(1.63) (3) > DT(1.61) (4)	0.042**
Multiplication strategy: Increasing 150% of instances in each class label
DRSA (1.71) (1) > RSA (1.68) (2) > CART (1.65) (3) > DT(1,62) (4)	0.037**
Multiplication strategy: Increasing 200% of instances in each class label
DRSA (1.73) (1) > RSA (1.70) (2) = CART(1.69) (2) > DT(1.66) (4)	0.021**
1.54 K = 5 (We assume this problem is a five-class label classification task)
Multiplication strategy: Increasing 50% of instances in each class label
DRSA (1.56) (1) > RSA (1.53) (2) > CART(1.50) (3) = DT(1.49) (3)	0.041**
Multiplication strategy: Increasing 100% of instances in each class label
DRSA (1.61) (1) > RSA (1.58) (2) = DT(1.57) (2) > CART(1.53) (4)	0.040**
Multiplication strategy: Increasing 150% of instances in each class label
DRSA (1.63) (1) > RSA (1.60) (2) > CART(1.57) (3) > DT(1.55) (4)	0.037**
Multiplication strategy: Increasing 200% of instances in each class label
DRSA (1.66) (1) > RSA (1.63) (2) > CART(1.60) (3) > DT(1.58) (4)	0.042**

** and *** denote significance at the 5%, and 1% level of significance, respectively

Back to article page