Mistake Master
Mistake Master · AP Precalculus · Unit 2 · Step-Through Animation
Fitting Is Cheap; Validation Is Earned
You'll learnto test which family a data set really supports, using two tests that can actually fail.
Any model can be forced through a point or two, and that is fitting rather than validating. Two tests that can fail decide which family deserves the data: the rate signature hiding in the table, and the shape the residuals make when you keep them in order. Watch three candidates compete over one data set, and watch a near-perfect fit turn out to say nothing at all about the far end of the graph. Twelve short steps; press Next for each one.
the rate signature in the table · the shape of the residuals
Before you start
What you're looking at
One data set with three candidate curves through it, tables of differences and ratios, and the leftover gaps plotted in order.
The question
Which family does this data actually support, and how far can the answer be trusted?