Demonstrating the practical aspects of data science
Alinta Energy is one of Australia’s largest suppliers of gas and electricity, serving over one million customers and employing over 800 people across Australia and New Zealand. Alinta Energy owns and operates power stations across Australian and New Zealand, and operates distribution infrastructure across Australia, delivering 3,000MW of power to Australian homes and businesses.
Alinta Energy plays a major part in developing a sustainable energy future for Australia, underwriting or investing directly in 1,500MW of large scalable renewables by 2025, including solar, storage and wind technologies.
The proof of value stage is key to developing an understanding of business needs and how to determine the appropriate solution to address them.
Alinta Energy has an existing data platform powered by Microsoft Azure and Databricks, the Alinta Data Hub (ADH), and this project aimed to further demonstrate the capability of the ADH through building a challenger model that could run parallel to Alinta Energy’s existing demand forecast models.
The challenger model offers business leaders an opportunity to perform detailed scenario analysis across various state markets and enhance forecast accuracy through the addition of more data feeds to incorporate weather impacts.
“We have made a lot of investments in the data side of the business. A number of significant data migration initiatives were completed in the last three years, and we wanted to prove the value of the platform we had invested in, with a focus on its ability to support advanced analytics,” says Buzby Kuramoto, GM Enterprise Solutions Delivery at Alinta Energy.
“We wanted help from people who understood the tech and had some familiarity with demand forecasting. This was about challenging what we had previously developed and that is why we brought Ignite in to help."
The team needed to ensure that the challenger model worked independently for each state and that they could train both states at the same time, minimising errors. They also needed to make the model customisable, so that users were able to easily create comparative scenario analysis.
Key to proving the value of the solution would be that the model could easily integrate into Alinta’s existing data science methodology and harness the power of existing technologies of Microsoft Azure and Databricks.
We run with a lean team at Alinta and having the bandwidth to be able to focus on this was key, so it was great to have help from an experienced team to drive the project. It is hard to find people that have a combination of the technical skills and a good sense of what a generator/retailer needs and Ignite’s deep experience in the Australian energy industry meant that we could rely on them for this.
The way that Ignite engages with business users and constantly evaluates their performance, at the end in particular, is probably one of their strongest points. They are always trying to improve the way that they deliver.
“The Challenger Model is now running in production and is used to challenge the assumptions of our base models. In general, forecast model performance is constantly changing and degrading, and the ability to spin up new models quickly enables us to evolve our forecasting as new technology and skillsets come along,” reflects Buzby.
“What we have now is something that can be run and supported internally in our strategic data platform. It remains our challenger model and is something that we can share with a broader group of people in the organisations. We now have more options to expand our capability to meet our growing forecasting requirements.”
Prove the potential value in data science ideas.
To help build sponsor confidence in your data science use cases, Ignite’s Proof of Value delivers:
“Our internal Data and Analytics team has built up good knowledge of what our trading team is after with respect to forecasting. The team can now provide better input to forecasting and the proof of value process allowed Alinta to refine our processes for developing analytics solutions in a repeatable way,” says Buzby.
“We have focused on getting our data foundations right over the past few years and now we have been able to show the more advanced capabilities of the ADH platform.”
“From a trading perspective, we can now look at multiple forecast scenarios with different fuels and across different states, challenge assumptions and do more in-depth analysis.”
The team identified that the next steps they could undertake would be to incorporate extreme weather forecast scenarios so that analysts could understand the potential impacts to the trading portfolio. They also discussed whether they could look to expand the new challenger models into medium term forecasts.
“We will be looking to expand the coverage of the models with other fuels across other states. This engagement has gone a long way in proving the value of Alinta’s investment in a strategic data platform and at the same time helped us build internal capability to expand the application of this project.”
We believe the best results come when data enables people