Improving Boosted Generalized Additive Models with Random Forests: A Zoo Visitor Case Study for Smart Tourism

Abstract

Smart Tourism for the Industry 4.0 and post Covid-19 challenge needs explainable AI Algorithms adapted for the Volatility, Uncertainty, Complexity and Ambiguity (VUCA) World with smart (physical components, algorithms, and IoT/mobile connectivity) elements. This paper shows how boosted generalized additives models (GAM) and random forest can be used in conjunction to improve the prediction and model explainability at the same time. This is achieved by using the predictions of the random forest as an outcome of the boosted GAM. Boosted GAMs can not only improve the explainability of random forest, but the random forest can also improve the predictability of boosted GAMs for modeling zoo visitors. This approach also has desirable regularization properties, such as model sparsity of the boosted GAMs. In addition, the current state of the art is provided and a detailed description with descriptive analysis of a case study for zoo visitors. The procedure with integrated XAI techniques, like variable importance measures and partial effects, is explained. In the future, the proposed concept can be implemented also for other industries or as a general method of XAI