PREDICTION ANALYSIS + FINAL THOUGHTS

Below, you'll find an analysis of how well my program was able to predict each of the six races at the state meet. The first number listed is the average difference between runners' places and their projected places. The second number is the average difference in seconds between runners' times and their projected times. The fractions (x/10, x/15, x/25) represent how many of the top 10, 15, and 25 I was able to get correct. Below that, I describe which teams finished ahead/below projections.

(Scroll to bottom of the page for final thoughts)

A BOYS

5.76

40.27

Missed Individual Champion

8/10, 12/15, 23/25

Got Team Champion

Flipped 3 and 4 teams.

A GIRLS

5.05

36.24

Got Individual Champion

9/10, 11/15, 22/25

Got Team Champion

Flipped 6 and 7 teams. Predicted 8th didn't run full team.

AA BOYS

6.01

28.17

Got Individual Champion

8/10, 10/15, 23/25

Got Team Champion

Flipped 3 and 4 teams. Predicted 8th finished 6th. Predicted 6th-7th finished one spot lower accordingly.

AA GIRLS

6.03

39.18

Missed Individual Champion

7/10, 13/15, 22/25

All Teams Correct

AAA BOYS

6.90

28.00

Got Individual Champion

7/10, 12/15, 21/25

Got Team Champion

All Top 8 Teams Correct. Predicted 9th finished 12th. Predicted 10th-12th finished one spot ahead accordingly.

AAA GIRLS

6.80

44.42

Got Individual Champion

8/10, 14/15, 22/25

Got Team Champion

Flipped 2 and 3 teams. Flipped 4 and 5. Predicted 8th-12th all finished differently.

Overall, I think the program did fairly well for its first season. In most cases, man still eclipsed machine in terms of prediction accuracy, but there are definitely improvements to be had in the algorithm. If I get the opportunity to return next year, you can expect better accuracy and potentially some deeper analysis.

With that said, perfect accuracy isn't the goal, and will never be attainable. Echoing the words of RunWV Josh: "Rankings and predictions don't mean a thing. That's why we run the race". My main hope is that what I've created can serve to analyze performances in new creative ways, and maybe even stir up some extra incentive for people who know they can outperform what the predictions say.

I'd like to thank everyone who expressed interest in my program over the course of the season. Getting emails from people all across the state who were curious or impressed was very motivating for me, and made me feel like the project was worthwhile. I'd also like to express my gratitude for everyone who reported errors in my data, or course changes between seasons. With your help, I was able to fix errors that I never would've known of otherwise, making my predictions all the more accurate. Most of all, I'd like to thank RunWV Josh for giving me the opportunity to present everything that I've worked on. His patience through every one of my errors, bad predictions and major data losses has been much appreciated. Last but not least, I'd like to give a big special thanks to the West Virginia running community as a whole. We're a very underappreciated state when it comes to cross country talent, but being able to experience the scene both first-hand as a runner and from a bird's eye view as an analyst, has convinced me that we have some of the most dedicated and talented runners in the nation, despite our small population. For everyone that's returning next year, I look forward to seeing what improvements that Track and the offseason bring. And for everyone that's graduating this year, I wish you good luck in whatever future running endeavors your life holds. Unless you're racing against me of course :)

As always, feel free to shoot me an email if you have any questions about the program.