Sequential Bayesian Method for Formulating Uncertainty in Sparse Data

To investigate uncertainty characteristics intrinsic in sparse data, a sequential Bayesian method was proposed to deal with this issue. In order to fully utilize information provided by original data, apriori distribution was calculated using apexes of the histogram of sparse data points for initial iteration. Mean square error was adopted as the assessing criterion of fitting fineness. Probabilistic measure entropy was calculated to describe how much information is used in both non-Bayesian and sequential Bayesian methods for comparison. Fitting examples demonstrate the capability of integrating uncertainty into the sparse data modeling process of the new method.