Some issues

[svilen]

Hi,
It is nice that you decided to open a forum for the users of your product. Since I am a frequent user of the program I've noticed some issues that I would like to post here. I hope that at least some of them could be implemented in the next patches of EViews (numbering doesn't mean importance):

1. INCLUDE function - EViews has a very nice feature to use parameters like {%something} but it doesn't work with all the functions. Such example is command INCLUDE, you cannot have something like:

Code: Select all

%path = "C:\.....\" include {%path}NameOfTheBatch.prg

2. database is slow - EViews imports data faster from xls and csv then its edb files. It was kind of odd to me. Is it because edb-s are sort of archives and the computer needs time to un-zip it (although my machine is quite fast)?

3. table as xls - EViews should be able to save tables as xls files (even without formating). Personally, I need this very much in my work.

4. Dated data report table - This is very convenient thing in mouse land but I would like to see all options that it has there implemented in the command line as well.

5. order pages - It would be nice to have this feature.

6. Select - It doesn't work properly when selecting upwards files with keyboard (you can select only 2 objects).

7. Unicode - You could put cyrillic names in Graphs, but in tables cyrillic characteristics don't work. It is important for me because of reporting which should be done in my own language. I am not sure that it is something about unicode. If it is possible for graphs why it is not possible for tables?

8. SETLINES bug - When using command setlines(area) with argument +o or +r it adds another column in the table on the right side of the area specified in brackets.

9. stochastic (Model Procs) - It doesn't work with options for s (d, m, s, b, a). I get an error message: "Error in Sample: Illegal date A in "DO_ MODEL.SOLVE"" (as a result of code: model.stochastic(s=a)).

10. Stacked charts - First, I should mention that I asked Gareth about this issue. His answer was this:

A stacked bar chart is supposed to "stack" the data one above the other (hence the name). Unfortunately Excel does it differently (in our opinion incorrectly) from most other statistical packages. EViews does not offer a way to do it in Excel's style.

Ok, from mathematical point of view this is right BUT if you have a tool you want it to do the job for you not the other way around. The main con of your approach is that you could have different graphs depending on the order of series (see the code below) and it becomes mess when you try to plot several series some of which with opposite singes of values. The main reason I want this graph to be done like in MS Excel is that I can easily see contributions of different components to something (which I my opinion is the main purpose of stacked charts). Please implement this type of chart in the future versions of your software and call it whatever you want if you think that "stack" is not the right name.

Code: Select all

close file wfcreate(wf=file, page=Q) q 2008:1 2008:4 genr ser1 = 2 genr ser2 = -2 group gp1 ser1 ser2 group gp2 ser2 ser1 graph gr1.bar(s) gp1 graph gr2.bar(s) gp2 show gr1 show gr2

[EViews Gareth]

Thanks for the long list of well thought out suggestions.

Most of them are already on our "to-do list", and some have already been implemented in 7.

Some are doubtful to be changed for a long time:


1. Changing the nature of the way INCLUDE works would be a major undertaking. I doubt this would be done unless/until we decide to give the programming language a complete overhaul.

2. This is just the nature of database files. Comparing to Excel or CSV isn't really fair, since they are flat files. You should really compare .WF1 saves/opens to Excel/CSV, and Access or ODB fetch/store to .EDB saves.

6. Could you give more information on this? I don't follow.

9. Are you sure your copy of EViews is up to date? The following code works perfectly for me:

Code: Select all

create u 100 series y=nrnd series x=nrnd equation e1.ls y c x e1.makemodel(m) m.solve(s=a)

[svilen]

I'm glad to hear that you have implemented some of my points in version 7 (btw is there any preliminary release date for the tentative/beta 7?)

On my issues:
1. If it is a major undertaking for INCLUDE then is it possible to introduce in 7 a new function that could work in a way I mentioned. It is good to have such option from organizational point of view. For example I have several prg-s that do the tables reporting, several prg-s that do the charts and some that do other things, and I prefer to keep them in separate subfolders to the folder where the main prg is placed. It is not a hard job to find/replace the paths in my main prg when I move the whole project to another location but still it is more convenient to have variable for the path.

6. I do it like this: select an object in workfile then on the keyboard I'm pressing SHIFT and UP arrow key. With that I could select only two objects. If I do the same with DOWN arrow key there is no such limit (or bug).

9. I didn't know that "a" could be used in SOLVE function (maybe you should mention it in the Command Reference). Instead I'm trying this and it gives me an error:

Code: Select all

create u 100 series y=nrnd series x=nrnd equation e1.ls y c x e1.makemodel(m) m.stochastic(s=a) m.solve(s=s)

11. One thing that I forgot to put in my previous post is the issue with your program editor. I appreciate very much that in 6 you introduced colored highlighting but why it is so slow? Our code is almost 1,500 lines and editing is fast enough but pasting is a kind of problem because it stops and it seems that it is checking whole the code to do the highlighting for a couple of seconds. So when I edit my code I prefer to use Notepad++ which is much faster then yours and I also made a schema for highlighting for it like the schema in your editor. I'm ok with that but I'll appreciate if you increase the speed of your editor so I could do the things in one place.

[EViews Gareth]

No tentative date yet.

1. Surely if they are sub folders off your main folder, then you don't need to refer to the whole path at all. Just do something like:
include .\subfolder1\program1
include .\subfolder2\program2

etc...

6. Never noticed that before. I'll see if we can change/fix it for 7.

9. Will look into it.

11. Yep, drives me nuts too. I've nagged the guy who programmed it so much, he's promised to get it fixed for 7.

[EViews Gareth]

With regards to 9., it is a documentation error.

The "S=" entry for stochastic, should actually be in solveopt (or solve). "S=" in stochastic is for the covariance matrix sample.

[svilen]

1. I tried it and it works. Then I don't need any change in INCLUDE command for the time being. Thanks for the nice solution.

[Tatha]

Hello,

I would like to second the request for an Excel-style stacked bar chart. In Eviews, negative bars are often overlapped and hidden by positive ones. The advantage of Eviews is that the last bar reaches the total of the components. This is crucial. But, for better visibility, the Excel approach is great. The way the Bank of England does these charts is to do an Excel style but with an additional 'line' showing the cumulative total (the line often passes through the middle of the bars, but this is no problem).

Tatha

[svilen]

12. import data from foreign file - It could be nice if EViews has options in command line alike that in mouseland for importing data. For example I would like to have an option to import fields like "Display Name", "Description", "Source" and "Units".

[ma79]

Hello,
I am an EViews ex-user and I switched long time ago. I wanted to help colleagues with some work in EViews
and found a problem with BAR CHARTS -- couldn't make contribution graphs wih +/- signs. I found your post (via Google) confirming
it is not possible.

So I wrote a routine that is supposed to do exactly what (I guess) you and I want -- plot bars +/- and when you sum it, it give you the total.
It is pretty simple, so modification is also simple.

Hope it helps. Just run draw_bar(series_group, range, "name_of_graph") and you get it.

I registered just to post the code for you guys, so you can reach me only by email in the code if needed.

Cheers

michal

[svilen]

Thank Michal, it works perfectly well!

[EViews Jason]

11. Slow programs when pasting has been fixed for some time. Installing the latest patch will speed things up significantly.