A new approach to data processing and uncertainty modeling based on the use of computational probabilistic analysis is considered. The basis of computational probabilistic analysis is numerical operations on probability density functions represented by piecewise polynomial functions. The problems of predicting the time series of distributions and estimating the probability densities of solutions of boundary value problems and systems of nonlinear equations with random coefficients are considered. Interval analysis, functional data analysis, symbolic data analysis and Monte Carlo method are currently used to study such data. A comparison of these approaches is given. Numerical examples show the effectiveness of the proposed approaches.