Predictive planning using
AI and machine learning

As we are always actively developing methods to enhance decision making and transparency for daily shipbuilding operations the capabilities of duration prediction were addressed within Floor2Plan recently. 

One such method is using artificial intelligence and machine learning to make predictions or learn about how and why deviations occur.

Insights from your historical data
We have recently developed a function that brings past experiences from the yard available for the future when setting up or maintaining shipbuilding projects.  

Predicting the time it takes to finish activities and assignments based on knowledge is therewith now possible within Floor2Plan. For instance a paint activity is set to take 5 days but we suspect from the computer model that (for the size of this engine room) it will take 8 days.   

A learning dataset 
As we retrieve knowledge from your dataset on how dependencies, sequences and activities work together and evolve we recognize numerous properties that together carry with them predictability.  

Currently we are training a model using machine learning. Using multiple tens of thousands of finished assignments from real data and their actual durations as examples. Some experiments have shown an accuracy of up to 70%, when grouping by duration in weeks, using over 40 fields of information ranging from involved disciplines, sequence as well as metadata from the components or parts consumed, like the weight or length of specific objects.

From explorations like these we do not only make available machine learning capabilities for your shipyard but we also learn where and how we can apply more of these principles in Floor2Plan. 

Meaningful impact in daily shipyard operations
But also we learn how we can empower our users with undiscovered clever insights by automatically highlighting specific areas that demand extra attention from project planners.