Achieving the first successful predictive model for maritime risk management
Achieving the first successful predictive model for maritime risk managementBy Ryan | October 14th, 2021
HiLo’s goal is to improve marine safety and save lives on the seas. To do so, we have broken new ground by developing the first successful predictive model for maritime risk management. This system identifies the real risks of incidents and highlights them to shipping companies. How did we do it?
What is predictive modelling?
Predictive modelling is a mathematical process that calculates the most likely future outcomes of situations. This is achieved by analysing past patterns, forecasting future results based on what has happened previously in the same situation. Historical data is collected, and then the analyst selects and trains statistical models.
Why has maritime predictive modelling not been done successfully before?
To make these models work, data needs to be accurate, granular and plentiful, and this is why nobody before HiLo created a successful predictive model in the maritime industry. Without full data from a large number of shipping companies, there isn’t enough information to generate accurate results.
Previous attempts have only been able to use incident reports, whereas HiLo has a huge amount of accurate data directly from vessels – and this database continues to grow.
What data is required for maritime predictive modelling?
Vessels are physically both complex and delicate, and keeping every aspect in the very best condition is a challenge:
- With so much to do, what do you need to focus on most?
- What faults pose the most risk, now and for the future?
- How can you prevent the most problematic issues?
Spotting the smallest and earliest warning signs is a skill in itself, but combining them in a network is key to making predictive modelling work. Take this example, about a single incident but with a range of useful information:
- Occurrence: Your maintenance log shows you need to repair a minor defect in the piping system ‘X’
- Fix: Product maintenance explains you need to procure exactly the right fittings to fix it properly
- Analysis: The incident report concludes that the pipe fault caused a small, seemingly, non-serious leak
Previous attempts at maritime safety predictive modelling, based only on the incident report, would raise a small warning that there was a non-urgent problem to fix. As a shipping company, with the incident management system as your only data source, you would come to the same conclusion which could be wrong.
All three pieces of data are correct, but individually they underestimate the true risk. Without all the required data, previous predictive models would have failed to identify the potential for a serious incident.
Putting all 3 aspects together, as can be achieved by HiLo’s model [for many such scenarios], shows the true picture: a critical risk of a larger spill with the potential of fire or explosion, where flammable liquids are concerned. Two vital points emerge:
- The immediate danger must be eliminated
- Vital improvements need to be made for safety practices going forward.
These findings allow you to prioritise and plan safety focus areas through a ‘risk ranking’ with exactly the right order and timings to prevent future incidents.
Creating an accurate maritime predictive model
The depth and quality of the data collected by HiLo combined with statistical and maritime subject matter expertise have resulted in the first truly accurate maritime predictive model, based around the following familiar equation:
Frequency x Severity x Probability = Risk
The frequency of events occurring is multiplied by the statistically calculated probability and severity to show the risk posed to the vessel and crew.
Our tool analyses hundreds of thousands of data points every year to identify the risks most likely to lead to an incident
How HiLo’s data and modelling makes your ships safer
What then makes HiLo’s maritime risk predictive modelling so effective?
In short, 4 aspects:
- We talk to mariners about their real-life experiences to record the frequency and nature of issues they experience every day
- Our risk severity and probability data is built on extensive expert research from Lloyds Register and Select Statistics
- This data is peer-reviewed by statistics experts at Imperial College London and the Alan Turing Institute.
- We have the largest and most accurate database ever compiled of full safety data from shipping companies, and it is growing month on month. The more data, the more accurate and far-reaching our analysis is.
Combined, this ensures we have the best statistical modelling in the maritime industry.
We are currently developing the first-ever predictive model to be used in the industry to quantify human behaviour, therefore adding another layer of information – human-related data – to reduce risks to the crew.
See how we achieve greater maritime safety
Watch our video for a deeper look inside HiLo’s predictive model:
Maritime risk management: Help us make the seas safer
The more companies who share their data with us, the more we can work together to reduce the risk of incidents that can impact the entire industry.
Our sole focus is to be the maritime industry’s data centre – making everyone safer onboard. All data is anonymous, so as a subscriber you can submit any and all data without fear of repercussion.
Try a demo today and increase your fleet safety
Request a free demo of the HiLo model today and see if it can work for you.
As well as receiving detailed, regular reports highlighting the Leading Events (LE) )or weak signals) that can cause serious incidents, you can take advantage of the experiences and expertise of industry peers through our knowledge-sharing forum CUPID.
Talk to us and help us to bring more seafarers home safely.