An Ontological Approach and a Meta-Algorithm for Data and Machine Learning Algorithms Analysis and Classification with Application in the Medical Domai

We introduce an algorithmic process, that we developed, which can be used for creating an Ontology and classification of various data sets and a meta-algorithm for finding the best suitable machine learning algorithms for a given data set to assist the knowledge discovery process. This new meta-algorithm involves a process for computing an encoding of a data set as a single numeric vector. A collection of training data sets from the medical domain is then clustered into clusters, to build a stratification of the data space. The most appropriate and successful machine learning techniques for each cluster are then derived by the meta-algorithm by determining the membership of the data set in the clustering. A testing set of medical data sets is used for testing the algorithm, including a data set for Rheumatoid Arthritis.