Sine Function with Alternating Peaks and Wavelengths

[hddd123456789]

Hi All,

I'm trying to find a sine function that models the cyclic and seasonal nature of retail sales across the years. I've tried several combinations including some basic attempts at complex plots of sin(x), but am now starting to wonder if there is even a simple solution to the problem. I've searched google for images of plots but maybe I'm not entering the right keywords?

Anyway I'm going to have to write out the description. The units right now don't matter since I can't even get the proper shape of the curve. I did a polynomial fit to average monthly data and got the fifth order polynomial that fits the data from x=1 to x=12 (for the months 1 to 12):

y = -1.225*x^5 + 40.785*x^4 - 479.82*x^3 + 2325.3*x^2 - 4029.5*x + 1554.433 view wolframalpha plot

I'm basically trying to find a sine function that repeats this basic pattern for every year since I need to then apply time-dependent attenuation and perhaps other functions to it. The closest I got was through complex plots like this:

y = (sin(pi*x-pi/2)) - re(sqrt(sin(pi/2*(x+2))))^2 view wolframalpha plot

Will greatly appreciate any help!

 

[Mute]

Have you tried fitting your data to a fourier series?

Fourier series are basically made to find all of the frequency components of periodic data.

 

[HallsofIvy]

There won't exit a single sine (or cosine) function that models that but a sum of such functions can (that is the "Fourier series" Mute mentions).

 

[hddd123456789]

Have you tried fitting your data to a fourier series?

Fourier series are basically made to find all of the frequency components of periodic data.

I haven't, how do you do that exactly? ;)

I'll just go ahead and assume that I'm in over my head with the actual math behind this for the moment. I am a programmer however, and found a nifty class that is supposed to perform a "Fast Fourier Transform" which calculates the "discrete Fourier transform". But this is apparently quite different than the Fourier series (from a google search).

Ok, some more searching around and I found this:

http://www.public.iastate.edu/~akmitra/aero361/design_web/crvft.html

at the bottom of which there is a link to an excel file that basically lays out all the calculations done. This seems to be what I need, and I would appreciate if someone could point out that I'm going in the right direction.

Many thanks!

 

[CellCoree]

I understand your frustration in trying to find a sine function that accurately models the cyclic and seasonal nature of retail sales. It is a complex problem and there may not be a simple solution. However, I would suggest looking into using Fourier analysis to decompose your data into its fundamental frequency components and then using those components to create a sine function that fits your data. This approach takes into account the different wavelengths and amplitudes of the peaks in your data, and may provide a more accurate representation of the seasonal patterns in retail sales. Additionally, you may want to consider incorporating other factors such as economic trends, consumer behavior, and marketing strategies into your model to further improve its accuracy. I wish you the best of luck in your research.