multivariable experimenta data : diaply with error bars and fitting
dokira
I have an experimental data which can be described by two variables (say x and y). I managed to display this data using gizmoplot by creating triplets using concatenate command. Here's what I did.
Concatenate { xwave, ywave, zwave }, tripletwave
NewGizmo
AppendToGizmo defaultScatter=tripletwave
1. My first questions is that, is there any way to display +-z error bars? For typical 2D plots, I used
ErrorBars ywave, Y wave=(ErrWave, ErrWave)
I'd like to show similar error bars in the scattered plot as well. The errors of zwave is stored in zwave_err.
2. Next thing I want to achieve is to fit the data wit a model function z=g(x,y) and display the fitted model function as a surface in the same gizmo plot. I was able to follow the multi-variable fitting part. But How can I display the the fitted model as a surfce in the same gizmo plot?
Thank you for your advice!!
When you do the fit, if you include the /D flag you will get a matrix wave fit_(name of z wave) with the model in it. That can be added to a gizmo plot as a surface.
John Weeks
WaveMetrics, Inc.
support@wavemetrics.com
April 6, 2011 at 12:59 pm - Permalink
There is more than one way to get error bars in Gizmo. First you will have to decide the dimensionality of your error bars (in standard XY graphs they are 1D). If you have 3D error data you might want to draw an ellipsoid at each scatter point. One way to do that is using a sphere for a marker and a scaling wave for the scatter markers. The math behind this is pretty messy so I am not sure it would be my first choice.
A simpler way may be to create another triplet wave for each error-bar dimension. For example, the wave for the x direction would have 3 rows for each scatter point. The first row contains the XYZ triplet corresponding to (x-dxl,y,z), the second row (x+dxh,y,z) and the third is all NaNs. Here dxl and dxh are the low and high side uncertainties. You then append a path object using the new wave and modify the "width" of the path to taste.
A.G.
WaveMetrics, Inc.
April 6, 2011 at 01:20 pm - Permalink
April 6, 2011 at 01:52 pm - Permalink