Plotting a 2D matrix

I have some data I would like to visualize, only I'm not sure which is the way to go about it. I want to take a 42x72 matrix called "data" of integers and plot:

f(i,j,data(i,j)), where i and j are index's of the data matrix. And data is the name of the matrix which holds the data. Or we can use a and b since this is matlab and we have problems with complex #.

I essentially want a 3D plot or a color plot which could help me visulize where the highest values are in the "data" matrix.

In addition, I have the curve fitting tool box. I wanted to know if some kind of function could be generated to describe the values in the matrix. So: f(i,j) = data(i,j). Can I find a function in mathmatical terms, rather than just having a big matrix, "data", that stores my points.

Thank you

Answer 1

[Sean de Wolski]

doc surf
doc mesh
doc plot3

to name a few visualization tools. You CAN fit a function to it, will it be a good fit? Not necessarily - that depends on your data/system.

[John]

thank you could you point out what document to look up for curve fitting?

[Sean de Wolski]

Open the curve fitting toolbox help and demos for how to use it. As for a GOOD fit - is there an analytical or empirical function that you data actually follows?

I would recommend reading John's reply nine here:

http://www.mathworks.com/matlabcentral/newsreader/view_thread/294201

[John]

Thank you. And, the data was generated in an anaylitical fashion using descision statements. It does follow a trend. Is there anyway I could email you the data file??

[Sean de Wolski]

Yes, but I may not look at it until tomorrow seeing as how it's almost five - happy hour :)

type

sftool

at the command line then use your data and analytical function.

[John]

It is definintaly non-linear. When I said analytical I mean it more from the method in how the data was generated. It's not random, but there is no analytical function.

[Sean de Wolski]

Then why do you want to fit an analytical function to it? What would you gain from having an analytical function (such as one generated by the SFT or a least squares fit)?

[John]

I wanted to see if I could fit the data to a polynomial. 5th order which max for me is not big enough

[John]

Thanks for the help

[Sean de Wolski]

You might as well fit (42*72)th order polynomial to it so it runs through every point! They'll both be wrong and for any real purpose, useless.

Other than seeing if you could fit a polynomial to it, why do you want to? What is your end goal I guess?

[John]

I am working in a memory limited enviroment<1KB. So having a large matrix or array that stored data points was not practical. I wanted to know if it was possible to generate the data on demand from a mathmatical function. This way I wouldn't have to store ever piece of data in an array.

[John]

The data represents probablity on a 2D map. So if I could generate the probablity just based on an equation, it would save memory.

[Sean de Wolski]

< 1Kb! Is there any option of getting more memory? If you just need to be able to generate points in this _exact_ space you could use a higher order polynomial. But it would be unstable for ANY other application.

Overall, I'd recommend getting more memory...