Diabetes Diagnosis using Clustering and Classification

In modern times a lot of data is being collected all over the world particularly in the field of medical sciences. All this data needs to be stored in an organised manner so that its retrieval is efficient enough. As the data increases it becomes more and more challenging to organise it. Clustering and classification are two major methods primarily used to organise the data. Clustering is the process of grouping similar objects and classification is the process of categorising the data. Usually these methods are used independently but in the proposed method we are combining both this methods to obtain better results. It uses patient's data where they are diagnosed for presence or absence of diabetes. The data is first normalized using Min-Max normalization method. Normalized data is then clustered using Bisecting K-means, K-Medoids and DBSCAN. The output obtained from clustering is then given to Naïve Bayes classifier. The output determines which of the combination for data processing the best is.