2.1. Data cleansing
The noises would significantly influence the ship collision risk assessment for the Singapore Strait. It is not appropriate to delete those records with noises in view of the real-time data integrality. We then propose a data cleansing procedure to eliminate the noises and update those inaccurate records. According to the Newton's laws of motion, the average speed can be calculated as the ratio of journey distance and travelling time. Therefore, the location records and the acceleration/deceleration abilities of vessels can be used to check whether the speed records are within the reasonable range or not. Correspondingly, the location data can be cleansed by using the updated speed data based on the same principle. The data pre-processing is illustrated in detail as follows.