Spectral Analysis*

For questions about EViews Add-ins available from the EViews Add-ins webpage. Note each add-in available on our webpage will have its own individual thread.

Moderators: EViews Gareth, EViews Moderator, EViews Esther

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: Spectral Analysis*

Postby startz » Tue Jun 09, 2015 3:50 pm

Note that this add-in is not an EViews product. It is an additional program generously made available by a user.

nicotrader
Posts: 63
Joined: Sat Jun 06, 2015 11:48 am

Re: Spectral Analysis*

Postby nicotrader » Tue Jun 09, 2015 9:39 pm

I did not want to be misunderstood, my note was not addressed to Nicolas user.
Sorry

NicolasR
Posts: 90
Joined: Mon Nov 04, 2013 6:22 pm
Location: Here

Re: Spectral Analysis*

Postby NicolasR » Fri Jun 12, 2015 9:50 am

If you are interested in a shot-run cycle you can estimate the spectrum by months or even by weeks as follows.

Code: Select all

wfcreate u 1 1250

for !i=1 to 96
   genr w!i=nrnd
   w!i.spectral(periodogram)
next


Where your week contains 1250 data points and w!i is the i -th week.

nicotrader
Posts: 63
Joined: Sat Jun 06, 2015 11:48 am

Re: Spectral Analysis*

Postby nicotrader » Sun Jun 14, 2015 1:16 pm

thanks Nicolas but analyze a time window on a weekly or monthly is not of any use in the financial field.
Another speech in the economic scenario....

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 13307
Joined: Tue Sep 16, 2008 5:38 pm

Re: Spectral Analysis*

Postby EViews Gareth » Sun Jun 14, 2015 1:55 pm

I don't really understand what you are requesting, and from whom.
Follow us on Twitter @IHSEViews

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: Spectral Analysis*

Postby trubador » Sun Jun 14, 2015 2:48 pm

nicotrader wrote:I think software should allow econometric analysis of time series high frequency, everything else is the past.

nicotrader wrote:thanks Nicolas but analyze a time window on a weekly or monthly is not of any use in the financial field.

These statements may reflect the personal views on the subject, but are simply wrong from an empirical perspective.
First of all, EViews allows the analysis of high frequency (HF) data. However, it does not have any built-in objects under such a specific category. You can always write your own procedures so as to suit your needs.
High frequency is not superior to other (lower) frequencies and vice versa. Each has its own merits and therefore is valuable and contributes to inferential process in the financial field.

nicotrader
Posts: 63
Joined: Sat Jun 06, 2015 11:48 am

Re: Spectral Analysis*

Postby nicotrader » Sun Jun 14, 2015 9:10 pm

EViews Gareth wrote:I don't really understand what you are requesting, and from whom.



It is very simple.
The user Nicolas has implemented the software with a great add-in, but as mentioned is not able to manage some thousands of observations.
I do not pretend to Nicolas revising add in for me but i demand that an econometric software that costs almost $ 2,000 to do it.
Simple right?

nicotrader
Posts: 63
Joined: Sat Jun 06, 2015 11:48 am

Re: Spectral Analysis*

Postby nicotrader » Sun Jun 14, 2015 9:12 pm

trubador wrote:
nicotrader wrote:I think software should allow econometric analysis of time series high frequency, everything else is the past.

nicotrader wrote:thanks Nicolas but analyze a time window on a weekly or monthly is not of any use in the financial field.

These statements may reflect the personal views on the subject, but are simply wrong from an empirical perspective.
First of all, EViews allows the analysis of high frequency (HF) data. However, it does not have any built-in objects under such a specific category. You can always write your own procedures so as to suit your needs.
High frequency is not superior to other (lower) frequencies and vice versa. Each has its own merits and therefore is valuable and contributes to inferential process in the financial field.



I'm sorry but it is not so ...
We just think of the memory feature long suffered by the HF data ....
and it could take dozens of examples to support my thesis.

nicotrader
Posts: 63
Joined: Sat Jun 06, 2015 11:48 am

Re: Spectral Analysis*

Postby nicotrader » Sun Jun 14, 2015 9:18 pm

I'm not here to make controversy!
This is the forum of a econometric software right?
Then carry my experience with the hope that we can improve this software.
Am i doing something wrong?
Or do we want to leave the complete supremacy to MATLAB?
I hate Matlab and i hope Eviews best.

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: Spectral Analysis*

Postby trubador » Mon Jun 15, 2015 12:07 am

nicotrader wrote:I'm not here to make controversy!
This is the forum of a econometric software right?
Then carry my experience with the hope that we can improve this software.
Am i doing something wrong?
Or do we want to leave the complete supremacy to MATLAB?
I hate Matlab and i hope Eviews best.

This is a discussion, not a controversy. And it is harmless.
The software (like all others) is in a constant improvement.
You are not doing anything wrong. You are just making very bold statements out of your own focus/experience, which I strongly disagree.
Traders are not the only group that makes use of this software. Or high frequency analysis tools are not the only tools that are used by analysts/experts.
Each software has its own merits as well. If it was not, then everyone would be using "the one". Unfortunately, this is a world of trade-offs.
The theory of high frequency analysis is still in its development stage. Therefore, you can always share your experience in the field, and point the developers in the right direction as to which technique has the utmost importance.
We, as users, are trying to do our best to help improve the software and you are very well welcome if you like to join the crowd.

ecofin
Posts: 182
Joined: Fri May 10, 2013 11:24 am

Re: Spectral Analysis*

Postby ecofin » Sun Dec 27, 2015 11:19 am

Dear Mr NicolasR,
this Add-in can make several smoothing window, can you add:
1-Tukey window;
2-Cumulative spectral distribution (http://www.stata.com/support/faqs/graph ... tribution/).
Kind Regards.

NicolasR
Posts: 90
Joined: Mon Nov 04, 2013 6:22 pm
Location: Here

Re: Spectral Analysis*

Postby NicolasR » Tue Dec 29, 2015 11:01 am

Hi,

The Hamming and Hann windows are special cases of the Tukey, do you mean a box to insert the parameter of the Tukey window?

The add-in already estimate the cumulative spectral distribution, is the distribution showed in the Normalized Integrated spectrum test which is the same as the wntestb command in stata. In the attatched workfile you can find the time series used in the estimates you showed, with the following command the test will be calculated (with the C.S.D) together with the 4 sinusoidal waves that compose the series and their periodogram.

Code: Select all

cos4.spectral(t,criteria=4,individual)


Regards,
Attachments
cos4.wf1
(12.28 KiB) Downloaded 420 times

ecofin
Posts: 182
Joined: Fri May 10, 2013 11:24 am

Re: Spectral Analysis*

Postby ecofin » Thu Jan 07, 2016 10:13 am

hi,
thanks a lot for the wokfile
1/ ok!
2/ how can make the cumulative spectral distribution for any series and show the graph?
3/ i use the Dow Jones Industrial Average (DJIA) daily , from 01/10/1928 to 24/01/2014 with 21426 obs, i generate a new series dly=dlog(y), my problem when i use the spectral analysis with the Hann or Hamming and the others window the error message appear out of memory...etc. but when i use this data with GAUSS 5 software it can make the spectrum with Tukey.
so my question: is the error that the window of Hann and Hamming not the same that Tukey in GAUSS 5? are there a solution to make spectrum with all windows with this Add-in.

Kind Regards.
Attachments
GAUSS 5.png
GAUSS 5.png (4.25 KiB) Viewed 17014 times
Hann_.png
Hann_.png (45.85 KiB) Viewed 17014 times
stocks.wf1
(1.37 MiB) Downloaded 404 times

NicolasR
Posts: 90
Joined: Mon Nov 04, 2013 6:22 pm
Location: Here

Re: Spectral Analysis*

Postby NicolasR » Sat Jan 09, 2016 2:42 pm

2. The spool "signal_tests" contains the graph and tabulate of the C.S.D you can extract any of them from the spool.

Code: Select all

dlg.spectral(t,criteria=4,individual)
signal_tests.extract(csd) graph_3

3. I can get the spectrum of the log returns using the Hamming window. The error of the out of memory appers because there are limits in the size of a matrix. I guess that i can calculate the spectrum and you do not because my machine has more memory (operative system of 64bits). I do not know if there is an option to increase the size of the matrix with which one works, Gareth this kind of option exists?

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 13307
Joined: Tue Sep 16, 2008 5:38 pm

Re: Spectral Analysis*

Postby EViews Gareth » Sat Jan 09, 2016 3:18 pm

No.
Follow us on Twitter @IHSEViews


Return to “Add-in Support”

Who is online

Users browsing this forum: No registered users and 20 guests