0
$\begingroup$

Is MATLAB better than R for time series analysis and forecasting or vice versa? What other software is considered best for time series analysis?

$\endgroup$
  • 3
    $\begingroup$ I think it is a possible duplicate, or at least there was a lot of discussion before stats.stackexchange.com/q/25/2645 In my ranking I would suggest open source R or Gretl for time series analysis, but it's just me. $\endgroup$ – Dmitrij Celov Oct 6 '11 at 9:54
  • 1
    $\begingroup$ If you wish to go beyond simple ARIMA models that assume no deterministic trends i.e. series of the form ,,,,,1,2,3,4 or level shifts ,,,,,1,1,1,1, or seasonal pulses ,,,,0,0,0,1,0,0,0,1,0,0,0,1 and pulse variables 0,0,0,0,1,0,0,0,0, AND that the parameters don't vary/change over time AND that the error variance doesn't change over time , you could be ok with the R or MATLAB. ARIMA models need to be robust to the ideas expressed here. $\endgroup$ – IrishStat Oct 6 '11 at 14:28
  • $\begingroup$ I did not know about Gretl. Just downloaded it. I hope it works well for environment data. $\endgroup$ – Bailey Oct 7 '11 at 3:21
  • $\begingroup$ I m searching for best time series analysis software, can any body suggest right choice, which one is like auto signal, or past, how to identify pattern in time series. more research data is at my blog homeoresearch.blogspot.com example data on the work. $\endgroup$ – user41231 Mar 3 '14 at 5:20
6
$\begingroup$

You can do roughly the same with the 2.

  • Matlab is maybe cleaner and easier to use as you have only one clean library of function for each task,
  • R is maybe more flexible as you have a LOT of packages that provide several methods for each analysis.
$\endgroup$
  • 3
    $\begingroup$ I would agree, and add that Matlab has a horrible syntax (if you can call it that) and costs a pretty penny. $\endgroup$ – Wayne Oct 6 '11 at 3:28
  • $\begingroup$ I have access to MATLAB at the uni. The main issue is the availability and ease of use of various methods. $\endgroup$ – Bailey Oct 6 '11 at 6:44
  • 1
    $\begingroup$ Would anyone consider Octave as a reliable clone of Matlab? $\endgroup$ – Owe Jessen Oct 6 '11 at 7:45
  • 1
    $\begingroup$ R is free and used more and more. I suggest to take this opportunity to test it. $\endgroup$ – RockScience Oct 6 '11 at 8:09
  • $\begingroup$ @Bailey Also if you have to deal with real precises dates the way Matlab handles dates is awfull as long as you have fractions of (Matlab) days that have infinite number of decimals ... $\endgroup$ – robin girard Jan 25 '12 at 8:33
4
$\begingroup$

Though there are lots of libraries in R for time series predictions, I frequently use "forecast" package. If my source data needs some cleaning or processing, I find python pandas very helpful, before porting data to R

$\endgroup$
2
$\begingroup$

I would suggest try Python which provides high productivity and it is free. They are some good libraries for Python. You can also have a look on a presentation Time Series with Python. FYI, Python+its packages (all free) could replace all Matlab+mathematica+... can do with additional advantages such as it is a very well know-highly appreciated fast growing programming language.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ As a forecasting tool, Python lacks packages compared to R. You can only do basic stuff (or you have to recode a lot of things) $\endgroup$ – RockScience Mar 4 '14 at 8:26
1
$\begingroup$

Stata and RATS are two other pieces of software that are popular for time-series analysis. OxMetrics also seems to be a popular choice.

Although I'm an R user, I must say that I am quite a fan of Stata for time-series analysis. The Stata [TS] manual is rather good. I prefer R, though, because it's more flexible. Of course, there is the Mata language in Stata, but I can't say I'm a fan of it.

I haven't used MATLAB, so I won't give an answer to the first question.

$\endgroup$
1
$\begingroup$

If you are interested in simple time series forecasting, i.e. without entering in the mathematical details of the process, you can have a look at SwiftForecast. It is quite new, but I find the idea of a "Google style" predictor quite interesting...

$\endgroup$
1
$\begingroup$

Regarding the second part of your question, I would like to introduce LDT. It is free. It provides automatic forecasting with stationary vector autoregressive (VAR) models and other types of time series analysis.

PS: I am the developer of this software.

| cite | improve this answer | |
$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.