Time series analysis using Mathcad!!!

[jkowalski]

I would like to ask if such a thing can be done using Mathcad?.
Or maybe you are somewhere examples of these solutions. It's mainly about the analysis and time series and forecast using moving average (MA), autoregressive (AR), and ARMA.

The new Mathematica 9 included:

• Support for scalar and vector moving average (MA), autoregressive (AR), and ARMA processes.
• Support for processes with polynomial and seasonal trends, as well as long memory time series models.
• Full support for simulation, estimation, and forecasting from time series models.
• Additional time series support for partial correlation function and spectral density.
• Flexible tests for unit root including augmented Dickey-Fuller and Phillips-Perron tests.
• Support for parametric stochastic differential equations (SDEs) such as Cox-Ingersoll-Ross.
• Support for scalar and vector Ito and Stratonovich processes defined by differential equations.
• Automatic conversion of parametric SDE processes to their Ito and Stratonovich forms.
• Support for Ito and Stratonovich SDE processes driven by other SDE processes.
• Support for nonlinear functions of Ito and Stratonovich process with automatic conversion to normal form.
• Extensive methods for simulation of SDE processes including Euler-Maruyama, stochastic Runge-Kutta, etc.
• Support for symbolically computing mean function, covariance function for time series, and SDE processes.

See web page.

http://www.wolfram.com/mathematica/new-in-9/time-series-and-stochastic-differential-equations/

[VladimirN]

Hi Jan,

Possible that my comments from this question will be useful for you:

http://communities.ptc.com/message/170265#170265

[jkowalski]

Hi Vladimir

Many thanks for your reply. Really appreciated.

The above is only maybe 50% of what it needs.

No ARMA(ARIMA)

No Dickey-Fuller and Phillips-Perron tests

No....

Maybe there are some add-ons,extrapacks,tutorials,workseets.?

JK.

[VladimirN]

You can bring this new and advanced capabilities for Mathcad in these groups:

"Enhancing Mathcad": http://communities.ptc.com/groups/enhancing-mathcad

"Mathcad Prime 3.0 Discussion": http://communities.ptc.com/groups/mathcad-prime-30-discussion

Or by creating a new "Idea" here (but only if you have an active/paid maintenance for Mathcad): http://communities.ptc.com/community/mathcad?view=idea

[jkowalski]

Many thanks for your reply, see what we can do in there..

[VladimirN]

No problem, Jan.

[Werner_E]

Maybe this could be of interest to you

http://isc.temple.edu/economics/notes/lectures_rev.html

especially chapter XVI

[VladimirN]

Werner Exinger wrote:

... http://isc.temple.edu/economics/notes/lectures_rev.html

especially chapter XVI...

Something is wrong with the indexes in the document ("armapq.mcd") for ARMA model: http://isc.temple.edu/economics/notes/timeseries/mcad_examp/armapq.mcd

Werner Exinger wrote:

http://isc.temple.edu/economics/notes/lectures_rev.html

Thank you for this link.

[Werner_E]

Something is wrong with the indexes in the document ("armapq.mcd") for ARMA model: http://isc.temple.edu/economics/notes/timeseries/mcad_examp/armapq.mcd

Yes, you're right. I don't think that Mathcad2000 (the version these files were made according to the site) was so much different. So I guess the file never worked. What is missing is the initialisation of the first x[i.

Changing as shown in the upper right seems to make it work:

[VladimirN]

Hmm... I opened this file (without modification) by using Mathcad Explorer 8 (wich you can find it here: http://communities.ptc.com/message/164132#164132) and it works fine.

[Werner_E]

Thats interesting. So vector initialisation was handled differently in those times.

The error you found is that Mathcad 15 does not allow assigning x[2 a value which is based on x[1 in a situation when only x[0 is set to 0. Mathcad seems to see x as a one element vector at that moment.

Mathcad 8 seems to even allow acces to x[-1 (assignment of x[2).

I was in the situation of changing a document created with an older Mathcad version to make it work with the newer one quite some times, but I didn't ran across this situation before. Nevertheless I would call it bad programming style. The things I had to change in other files were things which deal with functions of same name not behaving the way they did in older version. As an example the function arg(). In older versions arg(0) would return 0 while in newer Mathcad versions you get an error as (annoying, while mathematically correct) the argument is undefined in that situation.