To Avoid Overfitting Data, What Should be my Rule of Thumb for the Number of Data Points per Variable?

I am currently conducting some regression analysis to predict sales, and I currently have about 45 data points and have collected a lot of potential predictor variables. However, I am weary that I will overfit the data, so I'm wondering...is there a rule of thumb as to how many variables can be assigned given a number of data points, without danger of over-fitting the data? Thanks!

Amnesiac2011-04-18T10:58:34Z

Favorite Answer

Adjusted R^2 is a really useful tool in this situation. It is very similar to R^2 of simple OLS regressions, but it punishes the excessive use of predictor variables:

http://en.wikipedia.org/wiki/R-squared#Adjusted_R2

00

Related Questions

At what point in the relationship is it appropriate to ask someone about their birth data?How many points are thumbs down worth on Answers?Do you think we should be given bonus points for GIVING a thumbs up?How about we throw a thumbs up party ?!?Is there a rule to how much data you must have per dummy variable?

Trending Questions

Solve the equation sin(x+3)=sinx, 0=<x=<2pi to 4 decimal places. The product of the two solutions is:?7 answersHi, I am really stuck on this math question:?13 answersFind the area of the parallelogram if BC=16.?5 answersI need help finding volume?6 answersHow many metres are in a mile...?5 answers