Towards an overall picture of TU/e
How many first-year students will be starting their studies at TU/e on October the first? Useful information for a university that wishes to prepare itself at this early date. Up to now Education and Students Affairs (ESA) made this prognosis, but Merle Rodenburg, head of the Business Intelligence Cluster, says she can do a more accurate job by using more indicators. It’s a step towards an integral prognosis; call it the overall picture of TU/e. It should be finished by the end of this calendar year.
For over nine months TU/e staff members have been able to download reports and data through the portal of the Business Intelligence Cluster. The BI Cluster was designed in order to relieve staff in their search for data and to create a common frame of reference so that everyone works with the same figures. The optimal linking of data files from different domains has created more possibilities, says head of BI Cluster Merle Rodenburg.
Her group consists of staff from practically every supporting service and they spend part of their working hours making the prognoses. Rodenburg: “They go out and ask which input is needed to make a successful prognosis of something, and they also want to know what such a prognosis should look like.”
October 1 prognosis
The most recent development in the ever-expanding field of prognosis is the October-1-prognosis. How many first-year students will be enrolled in bachelor’s programs on October 1, 2019? “Up to now Education and Students Affairs (ESA) made that prognosis,” Rodenburg says, “but the Executive Board wants it to be more accurate, and updated every week. We can still make serious headway by entering data from previous years regarding pre enrolments and by taking into account the current situation." According to Rodenburg the initial idea was also to use machine learning, but at this moment there just isn't enough comparable historical data available to do that. She claims that her group’s October-1-prognosis will soon have a range of just 1 percent in terms of accuracy.
The prognosis went live yesterday and will adjust itself based on the most recent data during the coming weeks. The current prognosis shows that the two programs with fixed quotas, Mechanical Engineering and Built Environment, will certainly not reach their limits. The limit of first-year students for Mechanical Engineering is 330, but the prognosis indicates an expected influx of 176 students. An uncertainty factor of fifty students has to be taken into account though because at this point, it is still uncertain how many international students intend to enroll. Mechanical Engineering will be taught in English for the first time next year, and therefore there are no historical data as far as this area is concerned. Back in September 2018, 263 students enrolled.
Philip de Goey, dean of the Department ofMechanical Engineering, is not alarmed by the figures, “as long as they prove not to be structural. We expected a decline in the influx numbers this first year with a fixed quota, and we will be quite satisfied when the eventual number, including internationals, will be around 220. We certainly never expected to reach the limit of 330, and a few less students one a year isn’t a serious problem for us.”
Built Environment
The prognosis for Built Environment is currently at 174 first-year students, which is almost a hundred students less than the limit of 275. This academic year, the number of first-year students who started with their first year at the Department of the Built Environment was significantly higher at a total of 279. Jan van der Meulen, program director at the Department of the Built Environment, has reservations about the prognosis. “This is not in line with our expectations. A total of 390 candidates completed the selection procedure for their bachelor’s, 250 of them have accepted and a few more are still in the running. We expect to arrive at somewhere between 225 and 250 students eventually. So, there’s no reason for concern.”
Rodenburg says she understands Van der Meulen’s reaction: “The number of currently accepted enrollments is higher than our prognosis at several other programs as well, because our prognosis is based on enrollment patterns. Students will withdraw their enrollment during the coming months, for different reasons, despite having accepted a place in May. Because they didn’t pass their final exams, for instance, or because they enrolled in several programs at the same time and eventually opt for a different program or university.”
She says that the current prognosis is based on enrollment patterns and ratio’s between pre-enrollments in a certain week, and students who actually started on October 1. “That is a mathematical model on which our – lower – prognosis for Built Environment is based. It leads us to expect less students than the current number of 250 accepted enrollments. It’s still a prognosis, of course, and this is the best calculation we can make at this point, five months before October 1.”
The fact that the Department of the Built Environments has a fixed quota for the first time coming academic year has been taken into account by the model, Rodenburg says. “But we don’t have any historical data for this specific program in a year with a decentralized selection procedure yet. That is why it’s more difficult to make an exact prognosis for Built Environment than for other programs.”
Machine learning
“Eventually our method needs to result in an intelligent algorithm that allows us to prepare long-term prognoses for various aspects, such as influx, personnel, housing, finances. A combination of this has to lead to an integral prognosis, you could call it the overall picture of TU/e. We want to present it at the end of this year and it will then serve as a useful tool for the Executive Board to plan all the ambitions of Strategy 2030. But it’s not only useful to the board of course; staff members from every faculty or service will benefit from it in the future.”
An information session during which people can ask questions about this new prognosis will take place in Vertigo on Monday 20 May, between 14.00 and 15.00 hrs. The session is accessible to every user of the Business Intelligence Cluster and everyone who is interested in the BI portal, but it is specifically aimed at people who use the data made available in this report in their daily work, including program directors, heads of ESA, and educational policy staff.
Discussion