Dear Nallapaneni,

Yes, I do agree with Matteo that question is very generic and it is difficult to say anything. Nevertheless, I would like to give you some basic tips to deal with a clustering algorithm.

1. Try at least 3-4 clustering algorithm (K-mean, SOM, Hierarchical etc.). All have preferences depending on the purpose   (please do read the literature).

some goof literature here:

http://www.sciencedirect.com/science/article/pii/S0022169416301287

2. Make sure your resultant clusters are stable. People generally ignore this fact. Stable means, on running the clustering algorithm the resultant cluster should be same.

(cluster number may differ but the data in the cluster should be stable)

see that paper

http://www.sciencedirect.com/science/article/pii/S0022169416301287

3. You may compare two clustering algorithm

https://link.springer.com/article/10.1007/s11269-016-1428-1.

Hope this article will help you.

If you need MATLAB code used in this article, please do let me know.

Best, A

You need to first understand what sort of data you are trying to grouping. If your database already has class labels then instead of unsupervised clustering approach, you must use supervised classification approaches. I suggest you to use R coding for all sort of machine learning tasks. It is easy to learn and statistically it is a great tool. You may use GMM clustering for your work and then you can compare your result with well-known k-means and DBSCAN clustering. I hope GMM will give better result than others in terms of accuracy.

http://statweb.stanford.edu/~tibs/stat315a/LECTURES/em.pdf

Best of Luck