NIMAA: an R/CRAN package to accomplish NomInal data Mining AnAlysis

Abstract

AbstractSummaryNominal data is data that has been “labeled” and can be designated into a number of non-overlapping unordered groups. The analysis of this type of data is often trivial because it is not feasible to conduct extensive numerical methods on this type of data. Graphs or networks, on the other hand, are comprised of sets of nodes and edges that can also be considered as nominal variables. By integrating graph theory and data mining approaches, we offer the R package NIMAA to define a nominal data-mining pipeline to explore more information. Using nominal variables in a dataset, NIMAA provides functions for constructing weighted and unweighted bipartite graphs, analysing the similarity of labels in nominal variables, clustering labels or categories to super-labels, validating clustering results, predicting bipartite edges by missing weight imputation, and providing a variety of visualization tools. Here, we also indicated the application of nominal data mining in a biological dataset with well-riched nominal variables.AvailabilityNIMAA’s official release and the beta update are available on CRAN and Github, respectively. URLs: https://CRAN.R-project.org/package=NIMAA and https://github.com/jafarilab/NIMAAContactmohieddin.jafari@helsinki.fi; jing.tang@helisnki.fiContributionsMJ conceived the study and developed the models, MJ and CC adopted and implemented the methods, MM improved the methods, JT provided the funding, MJ, CC, MM and JT wrote the paper.