PacePlannr

Supporting information for my web app that provides personalized recommendation on pacing strategy for recreational marathoners.

Transcript

1. PacePlannr Personalized Pacing Strategy for Recreational Marathoners
2. THE PROBLEM…
3. OVERALL WORKFLOW     DATA
4. OVERALL WORKFLOW     DATA MODEL Fractional MLOGIT Elastic-net 72% accuracy
5. OVERALL WORKFLOW     DATA MODEL Fractional MLOGIT Elastic-net WEB APP 72% accuracy
6. Experienced runners run differently…
7. Experienced runners run differently…
8. My first marathon: 2012 Aquascaping
9. SUPPORTING SLIDES
10. Model Validation: k-fold MODEL VALIDATION k-fold cross validation (training set) % of (null) deviance explained = 72%
11. Model Validation: k-fold MODEL VALIDATION k-fold cross validation (training set) % of (null) deviance explained = 72% Evaluation on Test Dataset % of (null) deviance explained = 70%
12. OVERALL WORKFLOW     DATA MODEL WEB APP Fractional MLOGIT Elastic-net 72% accuracy
13. Top 10 finishers in the men’s NYC Marathon: 2010-2014 2011 2010 2014 2012 2013 Faster Finishes Distance
14. Top 10 finishers in the women’s NYC Marathon: 2010-2014 2011 2010 2014 2012 2013 Faster Finishes Distance
15. DATA SCRAPED 1,500,000 runners