The main idea of this project to provide an alternative method of analysis of K-means algorithm using wearable sensors. The presented work is a give the step of classification using K-means algorithm of data. It is well known classification approach for clustering in data representation. The analysis verifies and motivates the presented method which increases the advantage for intensity independent activity recognition. It is also present the design, implementation and testing of the K-means algorithm through complete experimentation by using wearable sensors. Finally performance analysis of the K-means algorithm on skin detection as well as features extracted discuss in detail.