Regression analysis for kg/hp vs. Ring laptimes

[jaiman]

Originally Posted by swamp2

R^2 is the most basic measure of the quality of the fit of the regression to the data (or data to regression). Agreed or disagree?

yes its the most basic, but that doesn't make it the most useful. Adjusted r squared is what we should be looking at as it will show which new variables are useful. Anyway, here's my final sheet with the three models.

[footie]

time kg/hp CD Skid Pad
Audi R8 484 3.798 0.38 0.98
Porsche 997 GT3 467 3.47 0.31 0.98
Nissan GT-R 458 3.594 0.27 0.99

All of the above show similar kg/hp and skip pad results but the CD is the only real thing that separates them and it's the only thing that seems to effect the lap times on the ring the most.

Clearly it looks like most of the times must be being made up during the long straights around the track where the superior CD is allowing a much higher top speed on these sections.

[Vince S.]

Love this thread. This is definitely one of the geeki-est threads.

Here's my contribution to comparing the cars. I have added a few more to Lucid's original list. But the dimensional vector for each car is still the same. The mapping here is based on similarity. Fed it through a proprietary algorithm I developed, here's the result. I need to include a few more dimensions for each vehicle (FWD/RWD, 2dr/4dr etc) so the map will be a little more meaningful.

With this map, you can get a better of who your neighbors are.

[swamp2]

Originally Posted by jaiman

yes its the most basic, but that doesn't make it the most useful. Adjusted r squared is what we should be looking at as it will show which new variables are useful. Anyway, here's my final sheet with the three models.

When ever I refer to R^2 I have always meant adjusted R^2 and have actually been reporting the adjusted R^2 values. However, with these data sets the difference between this two is farily insignificant.

I like your results. They again make my point: Skidpad results are much more important than Cd, Cd adds very little to the quality of the regression and you can see graphically the very low predictive capability of Cd.