UNC Charlotte Data Science Internship
Using machine learning on work request data.
imgId: 24

Survival probability of equipment over time shows that Dining equipment (red) takes a drop in survival after around 17 years.
imgId: 25

Classification of problem types accuracy results.
imgId: 26

Problem type classication results, the example shows both the original text the user submitted in the work request and the prediction of what problem category the work request should be classified as.
imgId: 27

Feature importance of the work request cost and time prediction models.
imgId: 28

Space Management dashboard showing employees of UNC Charlotte that do not have rooms assigned to them in the database. This helps with space utilitization and keeping information up to date.
Description
Had an internship with Facilities Management (FM) at the University of North Carolina at Charlotte, analyzing data from their work management system. Machine learning was applied to different areas of their business in order to improve operations. I developed models to predict the cost and hours for a work request which could help improve the scheduling of work, budgeting and providing better estimates to customers. Using natural language processing (NLP) I analyzed the work request descriptions in work requests and classified the problem type category. This can help improve how work requests are categorized and the speed at which they are routed to the appropriate shops to do the work. A big part of the work Facilities Management does involves installing, repairing and maintaining the equipment across campus. Analysis was done to study the lifespan of equipment and see if there are any patterns of when equipment will break down. This can help avoid outages and highlight where additional preventive maintenance scheduling needs to be implemented.
Skills Used
- Python
- XGBoost
- Word2Vec
- Deep Learning
- Tableau