Abstract: Errors of deformation monitoring data has great impact on the modeling of the data sequences. A new model called Event-Model, which changes the expression of the original observation data, was proposed. First, the deformation rate at each monitoring point in the original data was calculated, and then the rates were divided into several ranges. By converting the original observation data into events, the numerical data were divided into categorical data which were composed of the matrix, named Event Matrix in this paper, which could be fitted to the Cox regression model. Each group of categorical data contained certain ranges. On the basis of experiences and the actual situation such as the average displacement rate calculated previously, parameters in the “Event-Model” were adjusted to avoid or weaken the effect of observation errors. The mean square error of the observation data could be treated as adjusted parameters instead of unknown true errors in practice. The influence of monitoring data containing observation errors on the follow-up Cox regression model can be reduced to some extent and improve the accuracy of the calculation results.

Key words: observation errors, deformation monitoring, event model, Cox regression model