Transcript Artificial Neural Network System to Predict Golf Score on the PGA Tour

Artificial Neural Network
System to Predict Golf
Score on the PGA Tour
ECE 539 – Fall 2003
Final Project
Robert Steffes
ID: 901-685-8871
Idea
Use averages from seven of the major
shot categories to predict scoring average.
 Inputs include: Driving Distance, Driving
Accuracy (%), Greens in Regulation (%),
Putting Average, Birdie Average, Sand
Saves (%), and Putts per Round.
 The MLP is then tested using these inputs
and a scoring average is predicted.

Implementation
 Data
gathered from top 188 players
on the PGA Tour.
 Create training and testing files from
this data.
 Run through MLP with several tests
to get the optimum parameters:
3 Layers, 4 Hidden Neurons,
Learning Rate=0.1, Momentum=0.3,
1000 Epochs.
Results
 77%
average classification rate on
multiple tests run. Compare to 17%
random classification.
 No similar system implemented yet.
 Wide range of applications if used on
the PGA Tour.
Conclusion
 All
players have access to their shot
trends, averages, and statistics, but
it is virtually impossible to draw a
correlation just by looking at them.
 Potential applications beyond simply
forecasting a player’s score
– Eg. A player may hypothetically change
one of his statistics and see whether the
MLP predicts that that will change his
scoring average
Conclusion
 Professionals
looking for every
advantage they can get
– A system to analyze their statistics and
predict their scores could be extremely
valuable if utilized
I
would like to look into actually
developing a product from the
concept of this project in the future