A Robust Stochastic Method of Estimating the Transmission Potential of 2019-nCoV

The recent outbreak of a novel coronavirus (2019-nCoV) has quickly evolved into a global health crisis. The transmission potential of 2019-nCoV has been modelled and studied in several recent research works. The key factors such as the basic reproductive number, $R_{0}$, of the virus have been identified by fitting contagious disease spreading models to aggregated data. The data include the reported cases both within China and in closely connected cities over the world. In this paper, we study the transmission potential of 2019-nCoV from the perspective of the robustness of the statistical estimation, in light of varying data quality and timeliness in the initial stage of the outbreak. Sample consensus algorithm has been adopted to improve model fitting when outliers are present. The robust estimation enables us to identify two clusters of transmission models, both are of substantial concern, one with $R_0:8\sim14$, comparable to that of measles and the other dictates a large initial infected group.