Generating synthetic aviation safety data to resample or establish new datasets

Aviation safety data are limited in availability due to their confidential nature. Some aggregated overviews already exist but in order to effectively use the data, it is important to fill the gaps of their existing limitations. For some data, there are not enough data points in order to process them through advanced analysis. For other, only expert assumptions can be obtained. In both cases, these shortcomings can be addressed via proper data resampling or simulation where little effort can make the data suitable for various research and development initiatives. Examples of real aviation safety data made public are demonstrated together with key principles of how to perform their resampling. Then, for cases where only expert assumptions are available, general solution to the transformation of the assumptions into simulated data is introduced. The goal is to demonstrate how to transform accessible data or knowledge about aviation safety into data samples with sufficient granularity. The results provide general solution suitable not only for aviation safety data and knowledge, but also for similar transportation or high-risk industries related data issues, indicating that both the data resampling and simulation provide an option for generating datasets, which can be used for statistical inferential methods, linear regression modelling, recurrent analysis etc. Example of data resampling application is included in Aerospace Performance Factor calculation for years 2008 up to 2015.