University of Wisconsin Extended Campus is now Wisconsin Online Collaboratives! This name reflects the partnerships of the 13 universities within the Universities of Wisconsin–our state's premier system of public higher education. Through these partnerships we will continue to support online degrees, certificates and courses–along with support services to you.

Capstone Projects

Time Series Forecast of Single-Unit Residential Electricity Usage (A Case Study)

Program: Data Science Master's
Location: Not Specified (remote)
Student: Brett Kohout

My project created 24 time series forecast models to predict a consumer’s next month’s electric utility bill. The focus of the project was to develop a model(s) which consistently achieved a MAPE value of <= 15% and do so while leveraging data which could be considered minimally viable. Practically speaking this meant constructing univariate models relying solely on autoregression and multivariate models with features derived from the time component of the time series itself. This path was pursued due to all available research on the use case focusing solely on producing more and more accurate models, with no attention or analysis being paid to whether the model developed would be deployable and maintainable in a production environment. The best performing model was then analyzed for the length of training history required to achieve its peak performance and thus be operationalized.