An Evolutionary Algorithm for Automatic Recommendation of Clustering Methods and its Parameters

CARBALLIDO JESSICA A.; LATINI MACARENA; PONZONI IGNACIO; CECCHINI ROCÍO

Electronic Notes in Discrete Mathematics

ELSEVIER

Año: 2018 vol. 69 p. 229 - 236

1571-0653

One of the main problems being faced at the time of performing data clusteringconsists in the deteremination of the best clustering method together with deningthe ideal amount (k) of groups in which these data should be separated. In this paper,a preliminary approximation of a clustering recommender method is presented which,starting from a set of standardized data, suggests the best clustering strategy and alsoproposes an advisable k value. For this aim, the algorithm considers four indices forevaluating the nal structure of clusters: Dunn, Silhouette, Widest Gap and Entropy.The prototype is implemented as a Genetic Algorithm in which individuals arepossible congurations of the methods and their parameters. In this rst prototype,the algorithm suggests between four partitioning methods namely K-means, PAM,CLARA and, Fanny. Also, the best set of parameters to execute the suggestedmethod is obtained. The prototype was developed in an R environment, and itsndings could be corroborated as consistent when compared with a combination ofresults provided by other methods with similar objectives. The idea of this prototypeis to serve as the initial basis for a more complex framework that also incorporatesthe reduction of matrices with vast numbers of rows.