Comparison of K-means and DBSCAN for prediction determination of Down syndrome using prenatal test data.

Down syndrome is a condition in which a person has an excess of chromosomes. Babies are born with 46 chromosomes but Down syndrome babies have an extra copy of one of these chromosomes, chromosome 21. To predict a baby with Down syndrome, pregnant women can perform a series of prenatal tests. Clustering for the data in here is used to solve the problem that often occurs when determining whether Down syndrome occurs to group it into high or low risk because prenatal test data are invasive and most of the data will be immediately merged into one, therefore with accurate clustering predictions are needed so that patients can take the next step for the next test. In this study, we need k-means clustering and DBSCAN to compare which of the two unsupervised algorithms can predict the best clustering. The steps taken by involving prenatal test data attributes which will then be applied to the code of the two algorithms afterward are looking at the cluster level and grouping distance. The result required is a clustering without noise that has a small distance between the clusters that are carried out. From the results obtained, k-means that clustering is better than DBSCAN where the prediction level of clustering is better for determining Down syndrome even with a slightly high noise level which can only be removed by DBSCAN. [ABSTRACT FROM AUTHOR]