This is maybe a larger project. I would like us to be able handle (data-dependent) weights, affecting the contribution of the parametric and non-parametric part differently in different parts of x-space. An application of this could be to reduce the number of bumps in the tail: If the parametric density is small, give it more to say than the non-parametric part.
Maybe this can be used to make a method superior to that of Geenens for handling boundary bias.
This is maybe a larger project. I would like us to be able handle (data-dependent) weights, affecting the contribution of the parametric and non-parametric part differently in different parts of x-space. An application of this could be to reduce the number of bumps in the tail: If the parametric density is small, give it more to say than the non-parametric part.
Maybe this can be used to make a method superior to that of Geenens for handling boundary bias.