EViews 6 vs R 2.6.0

I previously mentioned that I had to use EViews this semester, instead of the already familiar R. Naturally, I wasn’t very happy, me being more used to R (and also, used to using R in my usual work). Then again, as they say, “never try, never know”. Grudgingly, I got EViews 6 last week, before semester started, and played around with it.

So, yesterday, I had my first econometrics tute (other people start their tutes next week, but we start ours this bloody week), and the tutor was going thru EViews. For fun, I took the sample data the tutor provided and worked out the same things in R (like calculating covariance and t-stats)… and so I thought, why not a comparison?

First Impressions
I initially found, in the boredom of my tutorial, that I could do the same things like plotting scatter plots way faster in R than in EViews… but it must be first noted, that I came from an IT background, so typing code is the norm for me.

I remember when I first opened up EViews (with their provided wfl file), I was overwhelmed by the amount of nonsense I saw on the screen. By nonsense, I mean a lot of GUI-ish stuff. On the other hand, when I opened R for the first time somewhere in the middle of last year, I was greeted by a familiar commandline. Ah, but that’s just the first impressions…

Under The Hood
I count myself generally objective when it comes to things. So I compared a few things between EViews and R (with all the statistical and graphical packages installed)

Working Speed
EViews is very graphical (yes, a commandline is given, but who the heck uses it), and things are generally done by clicking on buttons, be it generating graphs, or calculating heteroskadicity… you click buttons. In R, you type commands and codes. It’s a nobrainer who’s the winner in working speed – R. It’s faster to type code than to click around… worse still if you use a touchpad (I don’t bring my mouse everywhere)

But it must be noted that sometimes your data needs preparation, like transforming your data etc. EViews has some preset data transform methods, but the code needs to be manually typed in R. This may offset some of the click-vs-type working time. But all in all, R still remains a winner in productivity.

Speed
I must admit this – EViews is fast. REALLY fast. When I tested both programs with a sample database with 169 000 rows. EViews lagged a little when loading, but R didn’t. R simply… died at 100,000 points. Maybe its a setting error or something, but it was something cfgt and I have been working on, and the 100,000 row limit is a significant barrier for us both (though he reckons that there are about 42,000 data points that are way too insignificant to bother). EViews took all 169768 rows like a man, and didn’t lag one bit after.

The next task was to plot a probability density graph, and various relational graph. EViews again, fared much better, it’s fast, and R takes a lot of time to slowly plot the graphs.

Turns out though (we found out during summer), that our dataset needed to be split into 3 to make some meaningful estimates of things. R was good at splitting data. I’m not that good at EViews yet, but data manipulation-wise, I think R is faster (because of less clicks) and better (because I’m more familiar with it). EViews seems abit clunky in data manipulation, and I’m not talking about transforming data – I’m talking about cutting and splicing one series into 3 different serieses.

Features
Eviews wins hands down on this. One click, and I get all the statistics I need, from mean, standard devs, to kurtosis – all in a nice table (would you like a histogram with that?). In R, you’d have to type stuff like sd (x) to get the standard dev and stuff, and you have to do them one by one.

Plus, in EViews, I get to do all the tests I want to do (like testing for Normalcy, t-distribution tests, or even simple hypothesis tests etc) all with one click. Sure, you can type t.test (x…) but the data isn’t presented as nicely as EViews. Even easier and less complicated is the time-series stuff. I haven’t pored through the manual thoroughly yet, but I bet its gonna be easier than using R.

Price
Ah… R is free, and you can download R here. EViews 6 cost $39.95 for the student edition. Someone was very kind to me and donated a full Enterprise Edition + manuals ($645) to me because he was leaving the field of economics for another field. No feature/price benefit comparison here, because we know, trying to divide by 0 is oh shi-

Mac fanboy cfgt told me another reason not to use EViews – it doesn’t run on macs. Ah, but I’m a Windows user! LOL

Conclusion
In conclusion, I’m starting to really like EViews. Yes, clicking is still quite irritating to me, but it yields results quite well, and best of all, fast. There are somethings – simple things like plotting basic scatterplots that R is fast at (because of the working speed), but EViews processes data faster. R also allows fast manipulation of data, while clicking “edit +/-” in EViews is tedious and slow. Guess I’m going to use both, eh?