I would like to create a graphical model of a 7 variable problem in scilab
Problem: Create an address handling database system with incomplete data
The variables are as follows...
Primary address = [0,1]
Primary mailing address = [0,1]
Primary residential address = [0,1]
Primary billing address = [0,1]
Mailing address = [0,...,1]
Residential address = [0,...,1]
Billing address = [0,...,1]
Example: System asks for billing address and the following absolute values would be entered into the database
Primary address = 0
Primary mailing address = 0
Primary residential address = 0
Primary billing address = 1
Mailing address = 0.7
Residential address = 0.5
Billing address = 1.0
I would like to create a graphical model of how this multi-variable system will behave as certain values are held constant, basically creating a set of partial derivative graphs which will visually depict how the function behaves. Using plot3d I can plot a function of 2 variables, but is there a way to use color and time (animation) in scilab to extend this to two additional variables. This way I would only need to hold three variables constant.
For a research I did, I created a script that could do parallel space plots in SciLab for investigating the parameter space of algorithms. I based it on the papisr from Nelis Franken.
The image below is one of my plots. Each vertical line represents a parameter. I tested every combination of the five parameters within the ranges given. I qualified the results using a certain metric.
Each line represents a combination of the 5 parameters. I only plotted the top 500 results based on the metric. The gray lines are the results 500 up to 26, the darker the better.
The blue lines are the top 25 results and the red line is the best result.
In using this method, I was able to see certain trends and see which combinations of parameter settings worked best.