Role of clustering based on density to detect patterns of stock trading deviation

The pattern of deviation patterns can be identified from the results of cluster transactions and transactions that are transaction irregularities, will be detected. DBSCAN as a density-based clustering algorithm forms clusters that agglomerate and make it easier to detect unclustered data, which is considered as data noise (data outlier). The nature of density in the data clamping process will make it easier to determine noise data objects.The DBSCAN has two parameters, Eps and MinPts. The values entered in both parameters play a role in forming clusters. Stock trading transactions are stated as data objects to be clustered. The noise from clustering with DBSCAN shows outlier transactions, which have diferrent pattern with ordinary transactions. In the results of this clustering, the stock transaction pattern which includes outliers is obtained, marking the close occurs. This result can help to detect stock price manipulation in outlier transactions carried out by securities brokers.