Table 2.

Features and coefficients determined by the model using the multivariate adaptive regression splines (MARS) algorithm.

No.	Feature	Coefficient
1	(Intercept)	1.09
2	h(Cell_Size-2)	-0.32
3	h(2-Cell_Size)	-0.75
4	h(Cell_Size-3)	0.33
5	h(Bare_Nuclei-2)	-0.37
6	h(Bare_Nuclei-3)	0.42
7	h(Thickness-5)×h(Cell_Size-2)	0.01
8	h(5-Thickness)×h(Cell_Size-2)	0.02
9	h(Cell_Size-3)×h(2-Bare_Nuclei)	0.99
10	h(3-Cell_Size)×h(2-Bare_Nuclei)	-0.17
11	h(Cell_Size-2)×h(2-Bare_Nuclei)	-0.83
12	h(2-Epithelial_Size)×h(Bare_Nuclei-2)	0.21

h(Variable – Constant) are hinge functions representing knots the multivariate adaptive regression splines model identified to better fit the data. The results of the functions are the maximum of 0 and the difference between the variable and constant values. For instance, suppose Cell_Size is 3, then h(Cell_Size-2)=Max(0, 3-2)=1.