Using the Heimdall Machine Learning Platform to predict weekly retail shop sales

June 22, 2019

For smaller retailers that don’t have lots of data scientists with the skills to use low level command line tools, or for larger retailers where management or non data scientists want to quickly upload a flat csv to model builder. A 500 row model takes a few minutes to build and will give you a near real time prediction for future weekly store sales assuming you’ve taken recent data, and used feature selection tool to identify the predictive variables.

Our CTO and Co-Founder Justin Staines shares an example of predicting retail sales here:

Who is Yellow Submarine?

Yellow Submarine is an Explainable Machine Learning Self Service Platform.

Most Machine Learning models are opaque, non-intuitive and difficult for people to understand. We are different.

Our algorithms are all open box unlike competitive solutions that use deep learning algorithms such as neural networks which can become overfitted based on too much past historical data, or use outdated opaque off the shelf libraries such as R, Matlab, SAS or SPSS. All our algorithms have been coded from scratch minimising any opaqueness.

Our self service platform is data agnostic and doesn't require years of past historic data.

We use a combination of easily explainable GLM (Generalised Linear Models) and Clustering algorithms to provide inferences on near-real-time data.

The type of Machine Learning is known as Unsupervised ML.

Yellow Submarine is a privately owned company based in Bedfordshire. Its intellectual property is protected by patents GB1819646.9, GB1819645.1, GB1819644.4, GB1819643.6, GB1819642.8, GB1819641.0

Please visit for further information.

Predictive Crime product launched
New product release - Heimdall