Challenge – Build a sports performance calculation engine to power Circles - the Institute of Golf’s new player intelligence software. ​​

​

Solution – Our data science team developed bespoke machine learning models trained on ten years of player performance data (over 5.8 million shots). These models leveraged a range of modelling methodologies including Random Forest, Gradient Boosted Machines, Neural Networks and Autoregressive forecasting. ​

​

Outcomes - The models accurately determine characteristics of player performance and now benchmarks, predicts and provides personalised coaching recommendations in real time to golf players right around the world at www.drawmorecircles.com

Challenge – REINZ required an advanced computer-generated property valuation model to provide fast, accurate and automated property price guides for buyers, sellers and agents in the New Zealand market. ​

​

Solution – Our team of data scientists utilised state-of-the-art machine learning paradigms to develop a novel method of estimating the value of all residential properties. From urban apartments to rural lifestyle blocks, the model takes a range of market, location, sale and property data as inputs to produce up-to-date accurate estimates of a property’s current value.​

​

Outcomes - The model provides accurate results, far outperforming existing models in the market and is on par with cutting-edge literature in the field. The AVM now powers REINZ’s property price estimates and is publicly available at www.realestate.co.nz.

Challenge – Westpac needed an automated solution to improve its ability to manage branch network resourcing requirements. ​

​

Solution – We developed a machine learning model trained on three years of hourly transaction and branch demand patterns using two modelling algorithms (decomposition and temporal pattern isolation). The model is fully automated and self-evaluating, ingesting new data and re-training itself monthly.​

​

Outcomes – The model provides an accurate and robust two-month forecast of expected branch transaction volumes and FTE requirements at an hourly level. This has allowed Westpac to save time, optimise rostering and scheduling decisions and provide a much better customer experience across its retail branch network.