Since the 20 th century, the rail network's growth has been significantly increased worldwide to support passenger demands. One key to success is the safe service from the rail network, which shows a lower number than other public transportations. Nevertheless, many rail authorities have significantly increased safety level and reduced risks for a passenger. The causes of railway accident happened from various factors; however, primary accidents relating to infrastructure failures caused excessive damage to train and people's lives. All failures, which occurred in multi-parts of rail's infrastructure (i.e., roadbed, track, rail bridge), could be the primary causes of train collision and derailment. The overall goal of this study is to analysing uncertainties of railway accidents and, evaluating risk and resilience of rail's infrastructure after occurring an accident. The outcomes are expected to providing safety policies on the railway network. The research precisely conducts long-term global accident data sets, which related to infrastructure failures. The data sets are analysed by using Bayes' and decision tree methods through Python programming. One practical advantage of the study illustrates that the outcomes can apply to the railway networks' safety, reliability, and maintenance policies. Also, the research leads to sustainability surge railway's safety performances from avoiding infrastructure failures problems. As a result, the study reveals that the risk level from infrastructure failures shows at 'high risk' level that scored 18 of 32. Therefore, the research provides a practical recommendation to railway authorities to increase the infrastructure's safety level.