Algorithm 1.

Generating Intakes

Input:
	X = [x1, x2 . . ., xm], an array of vectors of features where xi = [f1, f2 . . ., famtFeat], a vector of features
	F = [f1, f2 . . ., fm], an array of food categories associated with a corresponding instance of X
	N, the minimum amount of intakes for each category
	R, the number of iterations before regenerating a new Random Forest
Output:
	Xnew: X concatenated with newly generated intake features
	Fnew: F concatenated with newly generated intake categories
	The length of each distinct type of Fnew is greater than or equal to N
1:	count = 0
2:	while ∃f(f ∈ F, ||f|| < N) do
3:	d ∈ F < N
4:	xd = x ∈ X | F = d
5:	for i = 1 to amtFeat do
6:	$I_{ne{w}_i}=\mathcal{N}\left(\overline{x_{d_i}},\sigma \left(x_{d_i}\right)\right)$
7:	if count mod R = 0 then
8:	Ftrain = [f : f ∈ F, ||f|| < N]
9:	Xtrain = x ∈ X | F ∈ Ftrain
10:	rf = Random Forest (Xtrain, Ftrain)
11:	else
12:	Use previous rf
13:	if rf.predict(Inew) = d then
14:	X.append(Inew)
15:	F.append(d)
16:	count+ = 1
17:	return X,F