Kernelized Fisher Discriminative Gradient Boost Big Data Classification for Disease Prediction

  • S.Midhun, Dr. A. Suhasini, Dr.A.Subitha

Abstract

            Big data is anarea used to analyze and extract the information for future use. The processing of huge amount of data is a difficult task and challenging one. Thesekinds of big data are applied in themedical fieldforearly diseaseprediction. The conventional algorithms process the big data since it failed to accurately learn the attributes from a raw dataset. A Kernelized Fisher Discriminative Stochastic Gradient Boost Data classifier (KFDSGBDC) model is introduced to enhance the disease prediction accuracy with big data. From big dataset, the number of attributes and data are collected. After that, the KFDSGBDC model performs the attribute collection by applying the Kernelized Fisher Discriminative Analysis.  This helps to select the pertinent attribute and discard there dundant attributes resulting in minimizes the disease prediction time. Secondly, the Stochastic Gradient Boost Data classifier model is employed to classify data with the pertinent attributes. The designed Classification technique constructs’ number of base classifiers that determine the input patient data as healthy or weak. Then the weak classifier results are combined into strong with the aid of gradient descent step-size function for reducing the training loss. This helps to decrease the false alarm rate and perform accurate disease prediction. An experiment is carried outusing various disease datasets. The obtained results show that the TSVR-CLPBC technique efficiently improves prediction accuracy and lesser complexity and false alarm rate than the state-of-the-art methods.

Published
2020-05-15
How to Cite
S.Midhun, Dr. A. Suhasini, Dr.A.Subitha. (2020). Kernelized Fisher Discriminative Gradient Boost Big Data Classification for Disease Prediction. International Journal of Advanced Science and Technology, 29(9s), 3234 - 3252. Retrieved from https://sersc.org/journals/index.php/IJAST/article/view/15883