Various medical sciences, educational, scientific research works are depends upon data clustering. Clustering is an application of data mining and knowledge processing which used find the data pattern using visual pattern analysis. Using this technique, data based on their attribute distance is mapped over space to find the pattern of data available. Clustering is type of unsupervised learning technique where no class level is available to utilize as feedback parameter for error correction. In this paper a study based on cluster analysis is provided. This includes study of k-mean clustering scheme, and outlier detection algorithm technique and performance enhancement using outlier technique, after concluding them we propose a geometric distance based clustering scheme. This technique is implemented using MATLAB simulator for experiment and academic research purpose, the implemented technique is works over numerical and the estimated results are provided in the further sections.