Line of Best Fit - Printable Version
+- Autograph Forums (http://www.autograph-maths.com/mybb)
+-- Forum: Using Autograph (/forumdisplay.php?fid=1)
+--- Forum: 2D Graphing Page (/forumdisplay.php?fid=4)
+--- Thread: Line of Best Fit (/showthread.php?tid=103)
Line of Best Fit - London - 28-11-2006 11:42 PM
I am having problems with creating a line of best fit and i have tried various polynomial orders with no success. The line is always crooked at the end. Any help will be much appreciated.
I have the following sets of data
Re: Line of Best Fit - Simon - 01-12-2006 05:16 PM
Thank you for contacting us with this interesting problem.
A polynomial is not always a good fit to data. In this case it looks like the data is decaying exponentially. To see this do the following:
1. Plot the function y = a*b^(-cx) + d
2. Select the function and click the "Animate Object" button on the Main Toolbar.
3. Adjust the parameters a, b, c, d until the function appears to fit the data well.
4. After you have tried adjusting the parameters, select the data set and the function and use the right-click option "Best fit to Data".
5. You should see that this curve fits very well.
Another very nice way of fitting the data is to first transform it and then try a (straight) line of best fit.
1. Double-click one of your data points
2. In the "Edit Data Set" dialogue box enter log(y) into the box next to "Scale-y".
3. Click "Scale-y" and "OK"
4. You will need to use the zooms to see the data properly.
5. You should notice that the transformed data now seem to follow a straight line.
Hope that helps.
Re: Line of Best Fit - London - 03-12-2006 09:12 PM
Thanks for you help. I have got the other method to work (the bottom one) but i am not able to see a line of best fit with this method but just the data arranged in a near straight line, which is good but i was hoping for the line to appear as well.
With the first method could you tell me roughly what values of a,b,c and d i should be using for this type of graph?
I am going way off and the line is not anywhere close to what it should be.
Re: Line of Best Fit - Simon - 04-12-2006 11:23 AM
In the first example if you select the data set and the line and use the right-click option "Best fit to Data". Then select the line and look for the values of a, b, c and d in the status bar.
In the second example you have only transformed the data you would still need to add a line of best fit. Select the data and use the right-click option y-on-x Regression Line.
Hope that helps,