Results toolbox

Restools

Current version is 1.7.0 (Feb. 2013).

This toolbox is a general toolbox for Matlab that should make it simpler to collect experimental results, annotate them, plot/print/show them in countless formats (like graphs, text tables, latex tables, html tables, bar graphs, surface graphs) and store them.

For this a new Matlab object 'results' is created. The idea of the results-object is:

  1. To extend the Matlab matrix with 'feature annotations', such that printing a matrix also shows the meaning of each row and column
  2. To be able to print the content of a matrix consistently in different formats (like text, latex, html, in a graph, a bar graph, etc)
  3. To extend the averaging over a dimension such that significantly different entries in a matrix are highlighted, and such that missing values (=NaN's) are handled correctly

The results-object should be as similar as possible to the standard Matlab matrix object ('double').

Download restools

Version restools 1.7.0 can be downloaded here.

Example

For the full details on how to use the toolbox, several example files are included in the toolbox. Please have a look at, for instance, the following script:

% This script should show how a results-object can be created, and how
% to make nice tables and figures form it.

% Generate (not completely random) results to show:
m = repmat((1:10)'/10,[1,4,5]);
res = m + 0.4*rand(10,4,5);

% Define the values of the different dimensions:
dim1 = {'cl1','cl2','cl3','cl4','cl5','cl6',...
   'cl7','cl8','cl9','cl10'};
dim2 = {'data1','iris','ionosphere','dat4'};
dim3 = (1:5)';

% Store it in a results object:
R = results(res,dim1,dim2,dim3);
% Set the dimension names:
R = setdimname(R,'classifiers','datasets','runs');
% Set the results name:
R = setname(R,'AUC performances');

% Average over the runs (dimension 3), and find the minima in each
% column (minimum over dimension 1):
T = average(10*R,3,'min1');
% And finally show the results, in text and html:
show(T)
show(T,'html'); show(T,'graph')

The commands in the last row of this script generates the following HTML-table and graph:

[restools graph]
AUC performances
classifiers datasets
data1 iris ionosphere dat4
cl1 3.0 (1.2) 1.9 (0.5) 3.4 (1.1) 3.7 (0.9)
cl2 4.8 (1.1) 4.4 (1.5) 3.6 (1.2) 3.0 (1.3)
cl3 4.4 (1.3) 5.5 (1.2) 5.2 (1.0) 5.2 (1.2)
cl4 6.7 (0.8) 5.4 (1.2) 6.3 (1.1) 6.1 (1.6)
cl5 7.1 (1.5) 6.6 (1.7) 7.3 (1.1) 6.7 (1.3)
cl6 7.5 (1.3) 6.9 (0.6) 8.1 (1.5) 8.2 (1.0)
cl7 9.0 (1.4) 9.8 (0.9) 8.7 (0.8) 8.8 (0.8)
cl8 10.6 (0.8) 10.4 (0.9) 9.5 (1.0) 10.4 (1.4)
cl9 11.3 (1.1) 11.1 (1.2) 11.0 (1.2) 11.3 (1.4)
cl10 12.5 (1.4) 11.7 (1.3) 11.2 (1.0) 11.5 (1.2)