We are on the verge of data explosion and there is an imminent need that some extensive measures should be taken to manage that gigantic bulk of data. One of the emerging and relying fields working on this problem is data mining. This thesis also serves the same purpose and introduces a novel technique to handle the data. First part of the thesis shows an existing technique being applied on large data and the later part proposes a novel technique. To show the authenticity of the proposed technique it is implemented on different types of data, compared with already renowned algorithm and in the end association rules are derived providing cross comparisons. Following the first part; depression is one of the emerging fatal diseases on world’s health care radar. This draft applies the data mining techniques such as association analysis and FP tree used altogether on depression database. The results from the analysis state the most common symptoms of depressed patients as well as provide an advantage of time and effort saving. In the later part a novel algorithm is proposed which imitates the concept of herding, i.e., dolphins catching their prey. This technique is a nature inspired method and introduces a new concept to cluster the data. It is an intelligent technique as all the data points are considered for every possible solution providing clustering and ideal centroids in a short time. This work presents the functioning of the algorithm on two different types of data sets that are artificially generated data sets and real medical data sets. It is believed that different results from various types of data can provide better diversity in results and also shows the validity of proposed algorithm. The results of the proposed algorithm are then compared with well-known fuzzy c-means (FCM) in terms of clusters formed presenting number of data items in each cluster, CPU processing time, simplicity and coverage and hence it has proved to be better than previous algorithm. After comparing it with the existing technique, association rules are derived from the unhandled raw data and the clustered data, i.e., before and after clustering the data. By doing so it will help us see the difference among results and also identify what new has been deduced in the results. Apart from comparison in data before and after clustering the same technique is applied on clusters formed to examine the bonding between inter clusters and intra clusters and it is likely to say that it fulfills the principle of clustering. The novel technique is manipulated and used every way so that no reservations are left to prove its efficiency and performance.