Volunteering with DataKind: Predicting Foster Care Case Complexity
Details
The Community Based Care of Central Florida (CBCCF) is the region’s leading child welfare organization, serving more than 3,000 kids in the foster care system each day. A major pain point for the organization is high social worker turnover, with an average of around 40% per year. The CBCCF partnered with DataKind and the Microsoft Cities Team to see if data science and machine learning could be used to reduce the caseworker churn rate.
Two projects were done to support caseworkers. The first involved a route optimization app that takes in planned caseworker activities and schedules them in an efficient way, while being flexible enough to take into account potential constraints the caseworker could have. The second project was a model was to quantify the predicted workload of each foster youth case.
This will be an interactive case study discussion (mainly on the second project) on the decisions made throughout the project by the data science team as we discuss the project from start to finish.