Benefit Analysis of Low-Carbon Policy Mix Innovation Based on Consumer Perspective in Smart City

Algorithm 2

BP-AdaBoost adaptive boosting algorithm.

Input: the training set {xs, ys, zs}, the test set {xd, yd, zd}. Initialize weight k. Calculate sample weight .
Output: predicting the carbon emissions of private cars {y}
(1)	k = 10.
(2)	For i = 1, 2, …, K do
(3)	Set the number of iterations e, set learning rate l, weak predictor training.
(4)	Calculate weak predictor predicting.
(5)	Calculate prediction error r, r = q − y.
(6)	Calculate test data predicting.
(7)	Adjust the weight.
(8)	For j = 1, 2, …, n do
(9)	If absolute value of error >0.2 then
(10)	New error = r +
(11)	New weight =   1.1
(12)	Else
(13)	New weight =
(14)	End if
(15)	End for
(16)	Calculate the weak predictor weights. Normalize weight.
(17)	End for
(18)	Obtain the carbon emissions of private cars {y}.