Once the data has been enriched with sufficient weather and ship context, it requires a specific mathematical treatment to make sense of it. Hydroscope leverages this highly contextualised data to select only the relevant information for each specific problem. Optionally a number of signal processing steps can help highlight and visualise the relevant information. Finally, the data can be aggregated with machine-learning grade interpolation methods to make the data interpretable and usable by the modern sailor.
data selection #
Because of the refinement of the contextual data added to the onboard dataset, it’s very simple to pinpoint specific data in Hydroscope. It’s possible to apply filters based on all the fields specified in the ship and weather context, leading to endless possibilities, for instance :
- select data where the 2022 main and 2023 J1 were rigged, and the hull was clean, and we had 12[%] camber on the main
- select data where at least 4 crew members where hiking, between 40[°] and 50[°] TWA, before the keel refit, with less than 1[°] of tilt on the keel
- select data where the ship displacement was less than 4.2 tonnes, with less than 0.1[kt] of current, more than 2[m] of significant wave height and less than 4[°] of foil rake
Each of these queries is just a few clicks away in Hydroscope, providing great agility to your data analysis.
signal processing #
Having selected the relevant data streams, some data requires mathematical transformations to reveal the useful information it holds. Especially when studying highly dynamic systems such as yachts, it’s usually helpful to either try to smoothen the dynamic effects with filtering and aggregation or instead to try to understand the dynamics at play in your data with specific timeseries analysis tools. This is why Hydroscope can perform any combination of operations (such as the ones listed below) on your data, to help extract the useful information rapidly.
- filtering
- statistical analysis
- resampling
- derivation
- integration
- fourrier analysis
- auto-correlation
- cross-correlation
interpolation #
Some important steps in the data analysis process, such as the computation of polars require the interpolation of data. Hydroscope uses state of the art machine-learning techniques to perform this challenging task which consists of “joining” the data “dots” together. The resulting ‘response surfaces’ not only fit the data very closely, but also compute the ‘uncertainty’ of their predictions, opening the way for informed decision-making.