[R] Stuck with FFT

Rthoughts seb.brockel at northampton.ac.uk
Tue Oct 28 11:04:41 CET 2008

EDIT: to assist you to assist me in turn, for which I am grateful, more
detail than the below needs to be provided.

I open tables and view them using Rcmdr so they become active datasets. I am
assuming that x101, x100, etc, are column headings because when the dataset
is edited they can be changed in numerical or non-numerical values. Lets
assume I change those column headings to VB! and V2, and I would like the
output to appear in V3. V2 contains the values to be fft'd. These are
practise values of 24h length sinusoids thus if a sinusoid is viewed as a
frequency value and the time-series are hourly values where periodicity, not
amplitude, counts, this is a 24hr frequency time-series fo hourly values.

Here's an example table (to keep it simple, there are 2 columns, V1 for the
24h frequency values, and V2 for fft output, I don not understand the fft
function enough yet if V1 should contain date/ time values and if V2 should
contain the array, and V3 the output values);

The table name, I assume, since it appears in the window border and taskbar,
is 'Dataset'.

   V1     V2
1 105
2 115
3 140
4 180
5 230
6 300
7 360
8 418
9 450
10 480
11 495
12 500
13 495
14 480
15 450
16 410
17 360
18 300
19 230
20 180
21 140
22 115
23 105
24 100
25 105
26 115
27 140
28 180
29 230
30 300
31 360
32 410
33 450
34 480
35 495
36 500
37 495
38 480
39 450
40 410
41 360
42 300
43 230
44 180
45 140
46 115
47 105
48 100

This is open as a table by view command. How do I go about fft'ing this?

If I can do this I can then in turn try to analyse my real radon data which
are hourly time-series (thus is a disctrete and periodic array). I have to
analyse it for periodicities that can be attributed to 24hr and 24.8hr tidal
cycles to confirm the substance of the first aim of my research.

Secondly, how can I view the output as a graph? I am expecting a frequency
domain from the time domain, thus a frequency of a certain amplitude should
be produced.

Thank you very much. I hope this clears the matter up a bit. If not, I am
happy to provide more details.

Rthoughts wrote:
> Dear all,
> Before I can get into serious Fourier analysis of Radon time-series I am
> practising with 24hour and 24.8hour sinusoids to assist with my
> interpretation of signals ittributed to tidal input to Radon time series.
> I am stuck. I have tried researching this to no avail. I am awating a book
> that should describe fourier transforms in detail and another one that
> should discuss the R application.
> I know what the command line is but I can't get it to refer to the array.
> What is 'z'? If I have a table open entitled '24hsinusoids' with date/
> time column by the hour from January 01 2008 00:00 to January 02 2011 
> 00:00 (X01.06.2008) (time column listed as X00:00) with corresponding
> sinudoidal signal value column (X100) to give a complete curve for 24hours
> spread as curves over the whole time-span, how can I fft from this table?
> fft(z, inverse = FALSE) gives me as; object 'z' not found.
> Please help and explain how R recognises the table and does fft's from it.
> The time-series is discrete and periodic (or should be, as I need to
> analyse my Radon time-series fro exactly this and I am very stuck now).
> Thank you with all my heart if you can help.

View this message in context: http://www.nabble.com/Stuck-with-FFT-tp20192942p20204314.html
Sent from the R help mailing list archive at Nabble.com.

More information about the R-help mailing list