Little explorations with HP calculators (no Prime)
|
04-03-2017, 03:03 PM
Post: #108
|
|||
|
|||
RE: Little explorations with the HP calculators
(04-03-2017 08:52 AM)pier4r Wrote: While I still do not know how you got the fitting formula, I understood that you did not try to correlate the 1st and 3rd column of this matrix, but the 1st and the 2nd. You are correct, and I hadn't noticed until now that I was using the wrong column! All due to a single digit typo in a regex expression I used to clean up the table (I didn't want to re-type all the values). I'm glad you took the time to double-check. Hopefully my mistake didn't cause too much confusion. My method of fitting the curve was far less scientific than may be appropriate, which is why I was concerned about its efficacy. In a nutshell, it was: - Used Excel to plot the data, observed that it was (roughly) a power curve reflected horizontally - Taking left side as first range, asked Excel to show the power trendline formula - Tweaked the formula slightly to more closely align the value at 3825 (to bring it closer to 425) - "Reflected" new x-values (7250-x), then plotted right half - Repeated the trendline-tweak for that range, with the same focus on the value at (new) x-value 3425 to re-align the peak (04-03-2017 08:52 AM)pier4r Wrote: That would be even better actually, then I have the underlying X and I can extrapolate the "variance modifier" as I want. Then perhaps my mistake was a "happy accident". At the least, it showed the performance increase of the overall process when replacing the table look-up with a computation. Glad you caught the mix-up when you did! |
|||
« Next Oldest | Next Newest »
|
User(s) browsing this thread: 1 Guest(s)