Density-based classification with the DENCLUE algorithm

Abstract

Classification of information is a vague and difficult to explore area of research, hence the emergence of grouping techniques, often referred to Clustering. It is necessary to differentiate between an unsupervised and a supervised classification. Clustering methods are numerous. Data partitioning and hierarchization push to use them in parametric form or not. Also, their use is influenced by algorithms of a probabilistic nature during the partitioning of data. The choice of a method depends on the result of the Clustering that we want to have. This work focuses on classification using the density-based spatial clustering of applications with noise (DBSCAN) and DENsity-based CLUstEring (DENCLUE) algorithm through an application made in csharp. Through the use of three databases which are the IRIS database, breast cancer wisconsin (diagnostic) data set and bank marketing data set, we show experimentally that the choice of the initial data parameters is important to accelerate the processing and can minimize the number of iterations to reduce the execution time of the application.