An Application of PCA on Uncertainty of Prediction

Abstract

Principal component analysis (PCA) has been widely used in many applications. In this paper, we present the problem of computational complexity in prediction, which increases as more input of predicting event’s information is provided. We use the information theory to show that the PCA method can be applied to reduce the computational complexity while maintaining the uncertainty level of the prediction. We show that the percentage increment of uncertainty is upper bounded by the percentage increment of complexity. We believe that the result of this study will be useful for constructing predictive models for various applications, which operate with high dimensionality of data.