How to create a "training sample"

[Patty Breslin]

I am doing work on a linear regression and the graduate student that is reviewing my work wants me to create a number of worksheets that have a randon sample of the data- say 60% of the rows. Then we fit our model to these different worksheets. Is there such a function or do I need code of some sort?

 

[Ruddles]

How long would it take you to do it manually, do you reckon?

 

[Patty Breslin]

Not that long, I suppose. If I wanted 5 samples- maybe an hour. But there is the question of how "random" that really is...

 

[Ruddles]

That's the issue - balancing the usefulness of the automated system against the time it takes to explain what you want and have it designed, written and tested.

Sometimes the boring way is the best way.