Algorithm 1: Savings Quantification
Input:	X: Training Data features: Selected set of features Y: Training Energy Consumption E: ECM Projects R: Reporting Period Data
Output:	Z: Savings dictionary
1:	model ← XGBoost()
2:	H ← initialize hyperparameters dictionary
3:	Z = {}
4:	for each ECM Ei ∈ E do
5:		X[Ei [‘name’]] ← Add new feature per ECM project and set default value to 0  R[Ei[‘name’]] ← Add new feature per ECM project
6:		for all data points Xj ∈ X do
7:			If Xj[‘date’] > Ei[‘date_start’] then
8:			Xj[Ei [‘name]] = 1
			end if
9:		end for
10:		features.add (Ei[‘name’])
11:	end for
12:	optimized_model = GridSearchCV (model, params = H, scoring = ‘rmse’)
13:	optimized_model.fit(X[features], Y)
14:	for  each ECM Ei ∈ E do
15:		R[Ei[‘name’]] ← Set feature value to 0 for all data points
16:		consumption_no_ecm = optimized_model.predict(R[features])
17:		R[Ei[‘name’]] ← Set feature value to 1 for all data points
18:		consumption_with_ecm = optimized_model.predict(R[features])
19:		savings = SUM (consumption_no_ecm-consumption_with_ecm)
20:		Z[Ei[‘name’]] = savings
21:	end for
22:	return Z