Page 1 of 1
3D plot
Posted: Sun Aug 29, 2010 9:43 am
by ivory4
Is there a way to plot 3D graph, given three time series?
Re: 3D plot
Posted: Sun Aug 29, 2010 4:58 pm
by TomDoan
Not with RATS. Depending upon what you're trying to do, you might be able to use GCONTOUR; that's often easier to read than an actual 3-D graph. There's so little actual call for 3-D graphs in econometrics that we've left that for programs that are more heavily used in engineering and the physical sciences.
Re: 3D plot
Posted: Sun Aug 29, 2010 9:40 pm
by ivory4
Like figure 3 of paper
http://faculty.washington.edu/ezivot/re ... vision.pdf
I notice that they draw graph from moments equations which are possible using contour graph.
But I tried to estimate two of three, given grid values of the other one. In this case, it seems that gcontour is not feasible.
Re: 3D plot
Posted: Mon Aug 30, 2010 8:31 am
by TomDoan
That's actually a good example of why 3D graphs are usually pointless. You really can't tell much about the pattern from the 3D graphs-it's the 2D projections that provide better information.
What the 3D graph is showing is a cloud of x,y,z points. The scatter graphs are just the same thing in two dimensions, like (x,y) and (y,z). Where there are many z's near a particular combinations of x,y, you get the darkening effect. Do SCATTER with STYLE=DOTS.