The Catholic Diocese of Maitland-Newcastle is halfway through a three year build of an advanced analytics and data science platform, through a partnership with Microsoft. The 173 year old diocese is starting with insights on its near 20,000 students across 59 schools, with hopes of using data to accelerate education outcomes.
The religious group believes computer science will help identify both gifted and struggling students by analysing data on their education history from various sources. According to those involved, the analytics platform is already providing insights and predictions that may otherwise be missed in isolation.
In conjunction with the Catholic Diocese of Maitland-Newcastle historical datasets – some 50 years worth – a range of data silos presents an integration challenge for the group’s head of data and analytics, Zanné van Wyk.
“We realised there will be data quality issues but it was much bigger than we anticipated,” says van Wyck, who leads the program. “And I realised for that reason we’ve actually got to invest in it.”
Speaking exclusively with Which-50, van Wyk explained the system is being built in a way so as business units outside IT can access predictive analytics, and when needed dive deeper into the information behind them.
More than dashboards
The Catholic Diocese of Maitland-Newcastle is using a Microsoft Azure based solution, leveraging cloud, data lake, analytics and governance tools from the tech giant to link various data silos to a self service predictive analytics platform.
“When the first draft of architecture was sent to me and was an architecture that would have delivered dashboards and reports … I sent it back to the drawing board,” van Wyk old Which-50. “I said ‘this is not going to deliver what we need. We need more than this, because these are our objectives.'”
The eventual platform still allows for dashboards and reports but, importantly van Wyk says, they are built based on the specific business questions at the time and don’t require a data scientist to access.
“If I do it for them I don’t have that business domain knowledge and neither does anyone in my team. We’re not in front of the coal face every day, as a principal or director, so they can tell much better stories [with the data].”
One and a half years in Van Wyck says the project has advanced well, in large part thanks to a leadership group that knows enough to admit when they don’t know enough.
“People who don’t know anything [about data science] – and they know they don’t know anything – are fine,” says van Wyck.
“But I’ve been in organisations where people know very little but they think they know a lot. Those are often very very difficult candidates to work with you implement these solutions.”
Outcomes and opt in
Van Wyck says the self service platform provides the bedrock for the diocese’ first major analytics endeavour, a “Student Intelligence” project.
“We’ve already started to work on the predictive model where we believe we can predict in primary school how students are going to do in HSC or in year 12 in high school using the attributes that we’ve built around a student.”
The attributes include the expected: classes, teachers, academic performance and attendance. But also add things like which year they started school and which schools they have previously attended.
“We have an understanding of which social economic status they came [from], which schools they came from, where they work, and then in addition to that we’ve got the external assessment [records].”
Van Wyck says the analytics help the diocese to better understand student growth and intervene when needed, for either struggling or excelling students.
“We put a face on a student. So I can look at every student and see every one, every individual that touched that student at a certain point in time. We understand the student from his or her context. And by doing that we are able to have that targeted intervention.”
Asked how the platform or algorithms avoid favouring or marginalising certain groups, van Wyck says the solution is more data and more context. She says the platform has already identified “gifted” children which had not performed especially well in traditional tests.
“We try to build as much context into student information as possible …Just because [students] don’t perform in a NAPLAN test doesn’t mean they’re not potentially gifted children.”
On security and privacy van Wyk says the platform has been built with strict governance controls. Teachers, for example, can only access information on their own classes while those in services departments sign a confidentiality agreement.
But a balance needs to be struck, van Wyk says, because the data’s utility comes from it being able to be shared.
Parents must also opt in to the data use which van Wyck says is always guided by a “data for good” principle.
The self service analytics platform is expected to be rolled out to other Catholic Diocese of Maitland-Newcastle communities over the next 18 months.