The Bayesian classifier is a fundamental classification technique. We also consider different concepts regarding Dimensionality Reduction techniques for retrieving lossless data. In this paper, we proposed a new architecture for pre-processing the data. Here we improved our Bayesian classifier to produce more accurate models with skewed distributions, data sets with missing information, and subsets of points having significant overlap with each other, which are known issues for clustering algorithms. so, we are interested in combining Dimensionality Reduction technique like PCA with Bayesian Classifiers to accelerate computations and evaluate complex mathematical equations. The proposed architecture in this project contains the following stages: pre-processing of input data, Naïve Bayesian classifier, Bayesian classifier, Principal component analysis, and database. Principal Component Analysis(PCA) is the process of reducing components by calculating Eigen values and Eigen Vectors. We consider two algorithms in this paper: Bayesian Classifier based on KMeans( BKM) and Naïve Bayesian Classifier Algorithm(NB).