When using "Proc/Make Graph" in a Model, the "Sample for Graph" defaults to the entire range of the Workfile. Since most models do not solve over the whole range, it would be nice if the default were to be the range over which the model was solved. This would avoid strange-looking end-points on the graphs. Currently it is necessary each time manually to edit the "Sample for Graph", which is painful. Alternatively, could the "Sample for Graph" remember the last setting? This would at least avoid repetitive manual editing of the "Sample for Graph".
Also, the graphs plotted via "Proc/Make Graph" do not appear to obey the Graphics Defaults on the main EViews Options tab; it appears to be necessary to change a graph's properties in its own window (eg, via the Template tab) each time if one wants the series' colours (and line widths, etc.) to obey the Global Defaults. It would be nice if the Global Defaults applied to these graphs also.
Regards
Donihue
Proc/Make Graph in Model (EViews 8)
Moderators: EViews Gareth, EViews Moderator
-
- EViews Developer
- Posts: 161
- Joined: Wed Sep 17, 2008 10:39 am
Re: Proc/Make Graph in Model (EViews 8)
I believe the sample is defaulting to the current workfile sample (not the entire range of the workfile).
Model solve also defaults to using the current workfile sample, so if you set the workfile sample to the range over which you forecast (by double clicking on 'sample' at the top of the workfile window or using the 'smpl' command), then the solve output and the graph should match up I think.
On the Graphic Defaults issue - it looks to me as though the global defaults are being used for most things but in some cases are being overridden. Line width in particular seems to be overridden for no good reason. I'll fix this in the next EViews 8 patch.
Color and pattern are also being deliberately modified in some cases. It's a bit hard to change this because people might be expecting the current behaviour now. Can you provide particular examples where you are unhappy with what EViews is doing with coloring?
Thanks,
Chris
Model solve also defaults to using the current workfile sample, so if you set the workfile sample to the range over which you forecast (by double clicking on 'sample' at the top of the workfile window or using the 'smpl' command), then the solve output and the graph should match up I think.
On the Graphic Defaults issue - it looks to me as though the global defaults are being used for most things but in some cases are being overridden. Line width in particular seems to be overridden for no good reason. I'll fix this in the next EViews 8 patch.
Color and pattern are also being deliberately modified in some cases. It's a bit hard to change this because people might be expecting the current behaviour now. Can you provide particular examples where you are unhappy with what EViews is doing with coloring?
Thanks,
Chris
Re: Proc/Make Graph in Model (EViews 8)
Thanks very much for your reply, Chris.
1. Sample: I take your point, but I still think it would be more "natural" for the graphs just to default to the "Sample (adjusted): " shown in the Model after solving. Your aproach would require an adjustment each time that "Sample (adjusted): " changed. That would be (probably many times) fewer than at present, but still not optimal.
2. Colours: well, for my models, the defaults make the Actual to be a thin blue line, the Baseline to be a similarly thin black line, and Scenario 1 to be an equally thin dark green line. At a quick glance, these are almost indistinguishable, especially when very close together. I personally change them to thicker blue, green and red lines respectively, as in the attached graph.
3. Since we are on the subject: as you can see from the attached graph, where the sample is longer than the solution (because of issue 1 above), the colour of the graph for the Actual is over-ridden at the beginning by the Baseline. This is inconsistent - the early part (where there is no model solution) should be blue, as is the end part, not green.
Regards
Donihue
1. Sample: I take your point, but I still think it would be more "natural" for the graphs just to default to the "Sample (adjusted): " shown in the Model after solving. Your aproach would require an adjustment each time that "Sample (adjusted): " changed. That would be (probably many times) fewer than at present, but still not optimal.
2. Colours: well, for my models, the defaults make the Actual to be a thin blue line, the Baseline to be a similarly thin black line, and Scenario 1 to be an equally thin dark green line. At a quick glance, these are almost indistinguishable, especially when very close together. I personally change them to thicker blue, green and red lines respectively, as in the attached graph.
3. Since we are on the subject: as you can see from the attached graph, where the sample is longer than the solution (because of issue 1 above), the colour of the graph for the Actual is over-ridden at the beginning by the Baseline. This is inconsistent - the early part (where there is no model solution) should be blue, as is the end part, not green.
Regards
Donihue
- Attachments
-
- Mocdified Graph
- forum1.jpg (23.7 KiB) Viewed 5501 times
Return to “Suggestions and Requests”
Who is online
Users browsing this forum: No registered users and 11 guests