Re: Constrain a polynomial trend line to intercept y-axis horizontally

John, I reckon I misquoted my dodgy formula which gave me the large numbers.
I'm sure I had the ^{...} on the x's. In any case. I've got it to work on my
full data set now, so that's my original question well answered.

For your curiosity, the x's were wind speed and the y's were noise level.
(LA90,10 mins) The guidance, dating back some years, talked of a using
polynomials of no more than 5th order to fit the data, but a recent article
(written by proper statisticians and not mere acousticians!) warned of the
dangers of overfitting the data - maybe 500 points with lots of scatter)
Using even a cubic means that at low wind speeds you are predicting the the
noise level flies up again and that high wind speeds the noise level could
drop off again, so they used to say, just do a piecewise solution and limit
the curve flat at the max and min. The latest advice is to ensuring that
your curve reaches a sensible level (15? 17? 20?) at zero windspeed by using
a even-number-powered polynomial and constrain the slope at/near x=0. 

Thanks for your help.

View this message in context:
Sent from the GnuMeric mailing list archive at

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]