ShinyAdaBoosting - An interactive dashboard to adaptative boosting algorithm

Abstract

Boosting methods are becoming more and more popular due their outstanding performance when compared with someo thers statistical learning techniques. The Adaptive Boosting, or simply AdaBoost, was one of the first boosting techinques developed, and consists, generally, in a linear combination of weak models (models that peform slightly better than a random guess) to built a strong classifier . The main purpose of this article was to built an interactive application, using the Shiny R package, that possibilities the user to apply the AdaBoost model to some datasets and observe the behaviour of the model performance concerning parameter’s variation, base learners, presence of noise and others aspects as evaluating aspects as overfitting, accuracy and computational time.