The focus of the workshop was to bring together data practitioners with an interest in predictive analytics.

Hard core data

“Data practitioners, with a focus on advanced, ‘hard core’ analytics that support predictive modelling or advanced statistical analyses, need a platform to share their experience and learnings”, said Lok Yiu, the Centre’s Senior Insights and Analytics Officer.

“We are moving towards a more proactive approach using machine and deep learning, so we need to work together more closely.

“While we are not necessarily using the same analytical tools and our focus is different, the methods we use and the analytical challenges we face every day are very similar.

“This work group provides a great opportunity to share our learnings and programming tricks, while allowing us the rare chance to have our work reviewed by fellow experts in this area,” Lok added.

Random forest method

Lok showed the workshop the random forest method for classification models, discussing how to use the method and how algorithm parameters can influence its performance.

Smaller groups then discussed potential issues that arise when building random forest models: how to deal with imbalanced data and what it means to measure the performance of a model from a business perspective.

Lok says more workshops are planned to help increase collaboration across government. “We are strengthening the potential use of our data via advanced analytical methods. This means a better understanding of the state services landscape.”

For further information or to express interest in attending please contact the Centre.