+- HP Forums (https://www.hpmuseum.org/forum)
+-- Forum: HP Calculators (and very old HP Computers) (/forum-3.html)
+--- Forum: HP Prime (/forum-5.html)
+--- Thread: Statistics 2Var Power curve fit (/thread-8328.html)
Statistics 2Var Power curve fit - KeithB - 05-10-2017 09:32 PM
I was playing with the curve fitting function/plotting and while trying the Power curve fit (bx^m) with the following data:
0 0
1 1
2 1.4
3 1.7
4 2
9 3
16 4
25 5
36 6
which should fit b=1, m= 0.5 pretty easily, I keep getting a "S1 - Error Insufficient statistics data"
The others work fine - though I cannot guarantee I have tried all of them...
What is wrong?
RE: Statistics 2Var Power curve fit - Joe Horn - 05-10-2017 10:14 PM
Suggestion: Try deleting the (0,0) data point. Most curve-fit software balks at negative and zero data for power fits, since it needs to take the log of the data.
RE: Statistics 2Var Power curve fit - KeithB - 05-10-2017 10:37 PM
Thanks.
I believe that was it. I reset the app and tried it again and it worked! I was about to chalk it up to a weird state issue, but added a 0,0 point and it fails.
I think the error can be a little more clear. I was thinking it was a "number of points based on the order of the polynomial" issue, so I kept adding points and no joy. Hard to believe that that error could be solved by *removing* points!
RE: Statistics 2Var Power curve fit - Joe Horn - 05-13-2017 02:55 AM
(05-10-2017 10:37 PM)KeithB Wrote: I think the error can be a little more clear.
Agreed. The HP 50g (with the same stats data as above, and set to the PWRFIT model) reports "LR Error: Invalid Σ Data LN(0)", which is a more helpful error message than Prime's.