Algorithm 1 Greedy Partitioning
1:	input: dataset $\mathcal{D}$ with features V
2:	output: the optimal partitioning ${\mathcal{T}}^{\ast }$
3:	$k\leftarrow 1,{\mathcal{T}}_1\leftarrow V$
4:	loop
5:	for all $T_i\in {\mathcal{T}}_k$ do
6:	$\{{\tilde{T}}_{i1},T_i\backslash {\tilde{T}}_{i1}\}\leftarrow \mathit{Qu}(\mathcal{D},T_i)$
7:	${\Delta }_i\leftarrow {\Delta }_{\text{total}}\left(\right\{{\tilde{T}}_{i1},T_i\backslash {\tilde{T}}_{i1}\}\mid {\mathcal{T}}_{\mathit{k}})$
8:	end for
9:	$j\leftarrow \arg {\max }_{1\le i\le k}{\Delta }_i$
10:	if Δj > 0 then
11:	${\mathcal{T}}_{\mathit{k}}\leftarrow {\mathcal{T}}_{\mathit{k}}\backslash \left\{T_j\right\}\cup \{T_{j1},T_{j2}\}$
12:	k ← k + 1
13:	else
14:	${\mathcal{T}}^{\ast }\leftarrow {\mathcal{T}}_k;\text{stop}$; stop
15:	end if
16:	end loop