Making Your Data Science Project Better by Thinking About the Use Case
Making Your Data Science Project Better by Thinking About the Use Case

Abstract: 

What makes a data science project “good”? Often, both aspiring data scientists and junior data scientists focus too much on the technical - using the most cutting-edge tool and focusing on the small details to eke out slight increases in model accuracy. While technical skills are important, it’s typically the use case of a project that dictates the quality of a project. A poorly thought out project with a great model is less impressive than a simple model that thoughtfully solves a difficult real problem. Whether working on a portfolio project or working on a project for employers, it’s important to have a clear stakeholder and problem in mind. This not only makes projects more realistic, it also gives guidance for the difficult decisions that often need to be made (e.g. about how to clean the data, which data to include, etc). This talk will give guidance on how to make sure your data science project connects with a real use-case, and how that will help make your projects better.

Bio: 

Tommy Blanchard is a Senior Data Scientist in the Health Division at Bose. He previously led the data science team at Fresenius Medical Care. He holds a PhD in Brain and Cognitive Sciences from the University of Rochester, and did his postdoctoral training at Harvard University.