SOFIA: An analytics recommendation system

Abstract

The SOFIA analytic recommender system helps non-expert users to select analytics (algorithms and their implementations) to fulfill a data mining task in an effective manner. The recommender is a novel framework that applies an ontology-driven approach to recommend a ranked list of analytics to fulfill a task based on the characteristics of the given dataset. SOFIA relies on an ontological representation of data science principles evolved by the data science community; it does not require training examples nor actual deployment of candidate analytics on given datasets.