Additional Computational Power

[trubador]

I have been using EViews for years now (more than a decade actually), and I have had the chance to observe and witness its evolution. Over these years, I have been able to perform difficult and cumbersome tasks with ease thanks to its user friendliness. Although I really believe EViews has an innovative, easy-to-use object-oriented interface, its solvers lack computational power. My humble view is that state-of-the-art computational algorithms are integral to today's world of quantitative research.

Powerful optimization methods (e.g. global and nonlinear solver engines) and evolutionary algorithms (e.g. genetic algorithms, neural networks) are now needed more than ever. Many other econometric/statistical software packages have implemented such tools and expanded their area of use. It would be great if QMS diminished (or alleviate) the need for us(ers) to switch to other softwares when the task at hand requires advanced techniques.

The LogL object, for instance, could be added new features and therefore is a good place to start. I am looking forward to such upgrades in my favorite econometric software.

Cheers for all those years...

[MarF]

I agree that new and more powerful optimization methods need to be added to the newer version.

[fmramos]

It's an interesting point. I'm not a so hard user, but during my MA classes, It was said several times about this issue and it was one of the reasons that we use other soft wares like OxMetrics package or Matlab, even in the econometrics of time series. Its useful to learn other packages, but EViews is pretty flexible to end user, less expensive than Matlab and they packages and, of course, there are advantages in specialization (eg. optimization of routines).