Nsupervised Iterative Clustering For High Dimensional Data for High Prediction Accuracy

The classification of the Imbalance data is a major research challenge in machine learning using parameter selection methods and data classification techniques. The challenges encountered in data learning and class imbalance learning are jointly dealt with the data streams that comprise distributions of very skewed class. The proposed idea in order to classify the majority class containing data points is to make a learning model which is predicted with wrong label by employing the existing data classification technique. Despite a considerable improvement in classifying the overfitting data in the cluster, it is still difficult to classify the high dimensional data. Hence, we propose a new technique termed as Unsupervised Iterative Clustering (UIC) to address the difficulties in handling high dimensional data. The method iterates selectively on new data points in the data streams to establish the cluster on each formed cluster using k Means Clustering Algorithm. We provide the experimental results on patient PIMA dataset. The outcomes reveal enhancement in the efficiency of the proposed method over the popular k means clustering algorithm.