Non-convex cost functionals in boosting algorithms and methods for panel selection

TL;DR

This paper introduces a non-convex cost functional in boosting algorithms, incorporating a correlation term to improve additive time series forecasting, supported by theoretical proofs and a new algorithm, with practical application to tourism forecasting.

Contribution

It presents a novel non-convex cost functional and an ArgMin algorithm for boosting, enhancing time series prediction accuracy.

Findings

Successful application to tourism forecast in Trentino

Theoretical proof of existence of minimizing sequences

Improved forecasting performance demonstrated

Abstract

In this document we propose a new improvement for boosting techniques as proposed in Friedman '99 by the use of non-convex cost functional. The idea is to introduce a correlation term to better deal with forecasting of additive time series. The problem is discussed in a theoretical way to prove the existence of minimizing sequence, and in a numerical way to propose a new "ArgMin" algorithm. The model has been used to perform the touristic presence forecast for the winter season 1999/2000 in Trentino (italian Alps).