Banca de QUALIFICAÇÃO: NICOLAS JACOBINO MARTINS

Cluster Analysis in Low Probability Databases

Mega ck-Means, ck-Means, databases with low probability, Data Clustering, Mega Sena, Fuzzy Logic

This study presents the development and evaluation of the "Mega ck-Means" algorithm as an innovative approach for data clustering in datasets with low probability events. The "Mega ck-Means" algorithm is designed based on a previously developed algorithm called "ck-means" and is capable of identifying specific patterns and clusters in datasets. ucted on traditional databases, comparing and validating its efficiency with two other well-known algorithms in the literature: k-means and FCM, using validation indices such as the DB index and Xie Beni. Additionally, tests will be carried out on databases with low probability events, showing which of the three algorithms has better efficiency in terms of clustering quality and compare with conventional database clusters.