A New Clustering Method Based on the Inversion Formula

Data clustering is one area of data mining that falls into the class unsupervised learning. Cluster analysis divides different classes by discovering internal structure set objects and their relationship. This paper presented a new density method based on modified inversion formula estimation. should allow to improve performance robustness k-means, Gaussian mixture model, other methods. The primary process proposed algorithm consists three main steps. Firstly, we initialized parameters generated T matrix. Secondly, estimated densities each point cluster. Third, updated mean, sigma, phi matrices. works quite well with datasets compared K-means, Mixture Model, Bayesian model. On hand, methods have limitations because this in current state cannot work higher-dimensional (d > 15). will be solved future versions detailed further work. Additionally, results, can see MIDEv2 best outliers all (0.5%, 1%, 2%, 4% outliers). interesting cluster even if do not outliers; most popular, for example, Iris dataset.