Could data help to predict school enrollment?
An accurate prediction of student enrollment is crucial for schools and for governments to provide an appropriate amount of resources, such as financial support , classrooms, teachers, accommodations and so on. For example, in September 2016, Redwood City, CA, found 130 fewer students are enrolled than enrollment projection, which caused $1.2 million loss of funding of state. Vice versa, we can anticipate what if a lot more students are enrolled than expected — expansion of class sizes, shortage of teachers and facilities.
As principals and officials all realized the importance of enrollment projection, a lot of consultants, statisticians and demographers are working hard on this for years after years. However, the accuracy is still in doubt. And immigrations makes prediction harder.
In Seattle, there is a Public Schools’ Enrollment Planning Department. They are now trying to predict enrollment down to grade level at each school. Brent Kroon, the head of that department said, “Kindergarten remains the hardest grade level to predict.” “In 2015, the overall enrollment projections have been within 0.5% of actual enrollment. 2014 it was off by 0.75%, caused by more kindergarteners entering the district than projected”
To predict kindergarten, they use birth data to infer the child headcount. And to predict grade 1–12, they use historical data and each October schools report current students enrollment by grades to the department. And enrollment counts are done every month, which means the data is quite updated at the department to do the next year projection. “Information from Choice School placement, school capacity changes, any boundary changes and new building openings are overlaid. From all this research and statistical analysis, important district decisions about staffing, building expansions, technology placement, transportation and future new school locations are made every day.”
From the experience in Seattle, there is a great amount of efforts taken in the enrollment projection. Governments need to collect and update various information and make the system dynamic. Before, the resources are distributed at a rough scale, now with dynamic data system, they could be more even. But the great efforts to build up and run a data system also could cost a lot, will it save more in return?
reference:
