help in interpreting a graph - URGENT

navneet1990

hello!

i obtained this graph after a study
could anyone please help me in interpreting it
like
i want to know whether it follows any definite pattern
any arithematic,geometric progression or a parabolic equation or anything??

thank you

nave

Attached Thumbnails

 

d_leet

It looks periodic, so you could express it as a sum of sine and cosine functions.

navneet1990

how do i do that??
is there anything being formed
could you state it

do you want more data
i could forward to you my data
some more information
would that help??

d_leet

Look up Fourier transforms.

J77

You could do some time-delay embedding.

Looks quasiperiodic from the limited data you have.

navneet1990

I can upload the actual data .
The data has a periodicity of 19 years. But in between it seems disjointed. The sequence is like this :
9
28
18
8
26
15
4
23
11
0
19
9
27
17
6
25
13
2
21
9
28
18
7
26
15
4
22
12
0
19
9
28
16
6
24
13
2
21
10
29
18

navneet1990

Uh.Huh. It is beyond me!!!

d_leet

What? You asked for suggestions without telling us anything about what you want to do, or the level of math you're comfortable with, and we've accordingly given you suggestions.

navneet1990

I am 12th grade student from India. I have obtained this data for a project . I am comfortable with trignometry , Calculus etc. But I can learn Fourier transforms if it helps me interpret the data effectively. Please give your suggestions. And thanks for the follow-up.

J77

Is that all the data you have?

What results do you want?

Graphically, look up the delay-embedding, in say, Yorke, Alligood and Yorke - Introduction to dynamical systems and chaos (I think that's the title.)

I don't think there's enough data for a spectral analysis using ffts...

navneet1990

I am studying the new moons dates while it transits a particular constellation from 2000-2040. I wish to know if an equation can be derived to arrive at the dates say 2100 for similar star positions using this formulae .

navneet1990

hope this helps

Attached Thumbnails

 

CRGreathouse

I recommend looking at the data modulo small integers -- say 10 through 60. See if anything comes up. That will help you fit it to a modified sine curve, or at least show potential problems of doing so.

navneet1990

could you elaborate more on that

CRGreathouse

Sure. For made-up data points (31, 1), (41, 10), (51, -5), (61, 1), (81, -5), (101, 10) you'll notice that when x is 1 mod 30 there's one value, when x is 11 mod 30 there's another, and when x is 21 mod 30 there's a third. Just transform your data so it's all in the form (a*n + b, y) with some fixed y in the suggested range (roughly, the expected period) and see how similar y values are across points with the same b values. If the data are periodic, there should be some choice of n that makes them line up perfectly; if they are near-periodic, then some choice of n should make them line up well (perhaps with a little bit of correlation with the a value, perhaps with just random variation).

navneet1990

if you use your theory in my graphwhat would be the results???