Table 5. Trained gradient boosting model parameters.

Features	Data subsets	Number of coefficients	Number of non-zero coefficients
2 features	Labs	480	148
4 features	Labs	960	365
8 features	Labs	1920	612
16 features	Labs	3840	452
32 features	Labs	7680	778
42 features	Labs	10080	732
	Labs+demo	10224	384
	Labs+demo+events	11568	374
	Labs+demo+events+triples	-	-
11 filtered features	Labs	2640	431
	Labs+demo	2784	330
	Labs+demo+events	4128	468
	Labs+demo+events+triples	95136	995