Method

Remember, one-dimensional CP profile for all possible values z of the explanatory variable j based the interest observation x_.

hjx_(z)f(x_j|=z).

vipjCP(x_) is the expected absolute deviation of the CP profile hjx_() from the model’s prediction f(x_), computed over the distribution gj(z) of the j-th explanatory variable.

vipjCP(x_)=R|hjx_(z)f(x_)|gj(z)dz=EXj{|hjx_(Xj)f(x_)|}.