LARKinfolab is involved in the ‘Green Wave project’ initiated by Trapps Wise. The ultimate goal of the project implies developing an app to optimize traffic flows on waterways. A ‘green wave’ on the waterways will guarantee a reduction in the use of fuel and a lower emission of harmful substances. LARKinfolab provides data science expertise in order to give an accurate speed advice for boats.
The challenge for LARKinfolab within the ‘Green Wave project’ is the following: there is a need for predictive modelling in order to determine an accurate Estimated Time of Arrival (ETA) for boats. Using this ETA, one can then calculate the recommended speed for each skipper in order to save fuel and reduce CO2 emission. Furthermore, LARKinfolab makes sure to clean up the real time data and properly implement this data, as well as we assist in creating a graphic representation of the Dutch inland waterways.
Real time data on all ships in the Netherlands is processed in cooperation with Trapps Wise. This data is stored in an Amazon environment and is pre-processed for predictive modelling. The model used in the project is a neural network, since this model fits best the acquired data. Our ultimate goal is to also include a multilinear regression model to control for the Neural Network.
The tools that are used to analyze the data are tools from the Amazon environment. We use Step Functions, Lambda, SageMaker and S3 Buckets. Moreover, all programming from our side is done in Python with its plethora of libraries.
In order to get the best ETA possible, – and by extension – a good speed advice, we are using a neural network. Together with Trapps Wise we have written the data in a form that allows us to start predicting ETA’s applying the neural network. The model currently uses historic data such as speed, ship type, location, path etc. However, eventually we strive to expand the data we are using by including weather data; historic, but also present weather data and forecasted weather data will be added.
