The RightShip Qi platform uses predictive analytics – a technology that learns from experience to predict future behaviour – to provide customers with the ability to better manage maritime risk.Read More
In Qi’s predictive model, vast quantities of ever-changing data are analysed by sophisticated algorithms to spot patterns and draw conclusions from data sets too large, diverse and dynamic for analysis with previous technology.
RightShip receives real-time data from over 50 sources, cleans it up, masters, and removes any duplicates to create our ‘golden record’, which predicts the likelihood that a vessel will have an incident in the next 12 months.
A vessel’s predictive risk rating in Qi takes into account up to nine key risk factors, which interrelate in a unique way in each individual circumstance. Up to six of the most significant risk factors will be considered in a vessel’s rating (including regulator risk, casualties risk, PSC indicator, builder risk, size, age, type, DOC PSC indicator and continuity risk) together with any of the four positive adjustment factors (an audit, classification, association or a RightShip Dry inspection) that are external to the predictive model.
The RightShip Qi predictive model has been independently verified to be working as intended by two recognised organisations.
We’re constantly testing the system predictions against real world outcomes. Incident rates on vessels vetted in 2017 show that 8.5% of vessels rated one star on the 1st January 2017 had an incident, as opposed to 1% of vessels rated 5 star. Likewise, 45 of every 1,000 one-star vessels received port state control detentions as opposed to only two of every 1,000 vessels rated 5 stars. One star vessels received 6.5 deficiencies per inspection compared with 1.4 deficiencies for every 5 star vessel inspected.
As is the case with all predictive models, RightShip Qi undergoes an adjustment every once in a while, known as a ‘retrain’.Read More
Retraining is a ‘business as usual’ process which helps to preserve the desirable features of the original model, increase performance, and remedy any shortcomings that have been identified while it’s been in use.
A retrain generally occurs every 4-6 months, when the model needs to adapt to changes in input data since the previous training. This ensures that the model is up-to-date and accurately reflects the data associated with the current fleet.
RightShip’s focus remains firmly on improving maritime safety standards, and Qi increasingly helps us target those vessels with a high likelihood of having an incident over the next 12 months. While retraining the model assists with this, the practical outcome of a retrain is that a minority of vessel star ratings may move (either up or down): however the vast majority will stay the same.
The following animation provides an overview of the retrain process. If you have specific questions relating to the retrain or changes to your vessel ratings, please contact firstname.lastname@example.org.