Visual Techniques to Compare Predictive Models

Predictive analysis is an important part of data analysis. Predictive models, based on Statistics or Machine Learning, are increasingly used to estimate, with a certain probability, future values of the variables that describe a phenomenon. Different models produce different results on a same dataset; thus, several models should be compared in order to identify the most suitable one. The paper is part of a larger research that aims at providing interactive visualizations that help the analysts to compare predictive models and to select the model that best fits the data. Specifically, two visualizations are presented, which support the analysts in performing some tasks of the Keim’s Visual Analytics Mantra.