Professional sports, like many other professions, require a variety of skills for success. That makes it difficult to evaluate and predict success. Fortunately, sports provides examples we can use to learn about modelling success because of the vast amount of data that are available. Here’s an example: What makes a golfer successful? The game of golf requires many skills. Putting well or hitting long drives will not, by itself, lead to success. Success in golf requires a combination of skills. That makes multiple regression a good candidate for modelling golf achievement. A number of internet sites post statistics for the current PGA players. We have data for 204 top players of 2006 in the file ch20_MCSP_Golfers.

These players all earned money on the tour, but they didn’t all play the same number of events. And the distribution of earnings is quite skewed. (Tiger Woods earned $662,000 per event. In second place, Jim Furyk earned only $300,000 per event. Median earnings per event were $36,600.) So it’s a good idea to take logs of Earnings/Event as the response variable.

The variables in the data file include:

Log$/E The logarithm of earnings per event.

GIR Greens in regulation. Percentage of holes played in which the ball is on the green with two or more strokes left for par.

Putts Average number of putts per hole in which the green was reached in regulation.

Save% Each time a golfer hits a bunker by the side of a green but needs only one or two additional shots to reach the hole, he’s credited with a save. This is the percentage of opportunities for saves that are realized.

DDist Average drive distance (yards). Measured as averages over pairs of drives in opposite directions (to account for wind).

DAcc Drive accuracy. Percentage of drives landing on the fairway.

Investigate these data. Find a regression model to predict golfers’ success (measured in log earnings per event). Write a report presenting your model, including an assessment of its limitations.