- Thread starter
- #1

- Thread starter catwalk
- Start date

- Thread starter
- #1

- Jan 30, 2018

- 485

- Thread starter
- #3

No, im not even sure what best fit means. I guess whatever is the simplest

- Jan 30, 2018

- 485

(Don't you just hate when they do that?)

Have you tried looking up "best fit" in your textbook or online?

- Thread starter
- #5

Haha yeah thats what happened.

(Don't you just hate when they do that?)

Have you tried looking up "best fit" in your textbook or online?

I've looked online but still don't know what to do

- Mar 1, 2012

- 716

You’ll just have to make your best eyeball estimate for the y-values.

Once you get a reasonable list of coordinates, do a google search for an online quadratic regression calculator and see what that can do for you.

- Jan 30, 2018

- 485

A more "sophisticated" method would the "least squares" method which is harder but uses all of the data. With y= a(x-h)^2+ k, for each data point (xi, yi) the "error" is yi- a(xi- h)^2- k, the difference of the data value and the computed value. For example, for the first ^data point, (0, 0) the "error" is 0- a(0- h)^2- k= -ah^2- k. For (7, 350) the "error" is 350- a(7- h)^2- k. If we were to simply add those we might have a negative error cancelling a positive error that we do not want to happen. So square each and then sum. To find the smallest possible error, take the derivative of that sum of squares with respect to a, h, and k and set the derivatives equal to 0. Again that gives three equations to solve for a, h, and k.