The computer should have a way of inputting the circumference of the wheel, usually in millimeters. The instructions for the computer will show how to calibrate it; even the cheapest $5 department-store computers I've bought had these instructions.

I actually rode a 20 mile course with a GPS on my handlebar and found that my computer was about 2% off, so I subtracted 2% from the setting in the computer. Do not do this if you're going over a course with lots of curves; the GPS will probably underestimate the distance a bit by cutting corners between samples.

I hope you're joking about where the magnet is making a difference. It's counting revolutions, the center of the wheel doesn't turn any less times per mile than the outside.

That said, you're better off with the magnet nearer the center of the wheel; if you put it near the outside you have two bad effects:

1) the magnet is moving past the sensor very fast at high speeds and the computer may start missing revolutions.
2) the magnet's weight can unbalance the wheel more at the outside than near the center.