An Efficient Bayesian Classifier in SQL with PCA

DS.Bhupal Naik, S.Deva kumar, G Sridhar Reddy

Abstract


As Bayesian classifier is a fundamental classification technique. We focus on an efficient Bayesian classifier programmed in sql
with PCA. We consider three classifiers: Naive Bayes and a classifier based on class decomposition using K-means clustering and Bayesian
classifier with dimensionality reduction technique PCA. We consider two complementary tasks: model computation and scoring a data set. We
introduce one of the dimensionality reduction techniques, Principal component analysis (PCA) to achieve more accuracy and reduction of
storage space. We study several layouts for tables and several indexing alternatives. We analyse how to transform equations into efficient SQL
queries. We also analyse how to calculate covariance matrix for PCA using SQL. We perform experiments on wbcancer and bscale datasets to
evaluate classification accuracy, query optimization & scalability. Our approach shows improvement in accuracy over the existing approach.

Keywords— Bayesian classification, principal component analysis, covariance matrix, Mean, variance.


Full Text:

PDF


DOI: https://doi.org/10.26483/ijarcs.v4i5.1676

Refbacks

  • There are currently no refbacks.




Copyright (c) 2016 International Journal of Advanced Research in Computer Science