There are many datasets in the world that are formed in such a format below:
var1 var2 ... varM res1 res2 ... resN
----------------------------------------------------------
v(1,1) v(1,2) ... v(1,M) r(1,1) r(1,2) ... r(1,N)
v(2,1) v(2,2) ... v(2,M) r(2,1) r(2,2) ... r(2,N)
...
v(K,1) v(K,2) ... v(K,M) r(K,1) r(K,2) ... r(K,N)
Which is a K(M+N)* matrix. Each line means one "experiment result" (possibly). This is more like a multi-input, multi-output blackbox, while the variables settled down, we can know the results:
var1...M |----------| res1...N ---------> | The BOX | --------> |----------|
Sometimes, we want to know how the variables influnce the results. But... The data is too huge, and I don't want to sort it, select data, draw graph each time to see the curves. So...
I decided to write this tool to draw all the curves for me.
Input of this drawer should be *.csv file currently.
Currently this tool only works under Linux.
- gnuplot
- convert
- eog
- Chart::Gnuplot
- Term::Menus
Please check --help.
Please take sas-15k-data.csv as am example. It's format Should be something like:
var1,var2,var3,...,varM,,res1,res2,...,resN
var1,var2,var3,...,varM,,res1,res2,...,resN
var1,var2,var3,...,varM,,res1,res2,...,resN
Here to seperate variables from test results, I used two commas ,, to seperate (or to say, add an empty column beteween the two blocks if you open it in an Excel application).