Nowadays, large datasets become main intentions of researchers in many areas. However, a challenge that still remains mainly unresolved is the lack of strategies used for analysing large time-series datasets in parallel. Therefore, this research aims to design a model of exponential smoothing working on parallel computing by using the bootstrap method. Three parts will be considered in the model: data pre-processing using the bootstrap methods, parallel exponential smoothing, and aggregation of results to be the final predicted values. To implement the processes, some packages available in the R environment such as "foreach", "forecast" and "doParallel" are utilised. R environment provides many packages for scientific computing, data analysis, time-series analysis and high performance computing. For testing and validating the proposed model and implementation, a case study in astronomy, i.e. the prediction of asteroid's orbital elements, was done. Moreover, a comparison and analysis with the results produced by algorithm of Regularized Mix Variable Symplectic 4 Yarkovsky Effect (RMVS4-YE) is also presented in this paper to provide a high level of confidence on the proposed model.