Algorithm 2.

Entropy based Subset Selection

visx[i] : no. of iterations row i was selected
iterations: no. of iterations user has gone through
impact[i][j]: no. of times coli is filled but colj is missing
entropyx[i]: entropy of row i
$entrop{y}_y^{rows}[i]$: entropy of column i for {rows}
I[x][i]: Indicator function = 1 if value is present in row x and column y, 0 otherwise {rows}
1:	procedure GENERATE_SUBSET(C)
2:	X = C.rows, Y = C.columns
3:	$temperature=\frac{\mathit{\text{italicized}}}{100}$
4:	for $i=1\to X.length$ do
5:	$score{s}_x[i]\leftarrow \frac{iterations}{vi{s}_x[i]}+entrop{y}_x[i]+{\mathit{missing}}_x[i]$
6:	end for
7:	$rows\leftarrow \text{select p rows with probability}=score{s}_x$
8:	$y\leftarrow$ select column to optimize for, with probability proportional to missing values
9:	for $i=1\to Y.length$ do
10:	$scores[i]\leftarrow sim[y][i].impact[y][i].{\sum }_{x\in rows}I[x][i]+entrop{y}_y^{rows}[i]\times temperature$
11:	end for
12:	columns = Top q indices from scoresy
13:	return $c=rows\cap columns$
14:	end procedure