Tutorials

Complete tutorials are at http://ginnet.gforge.inria.fr/tutorial_files.

Demonstration

A demonstration video of GINNet (commented in french) is available on your web site at http://ginnet.gforge.inria.fr/.

Tool tips

Beware that you can often get an help on any type of graphical item, just by letting the mouse cursor over it few seconds.
A description of this item will happen and look like the one below (in blue):
A screenshot of a tool tip in GINNet

How to quickly use a model to study your data:

First step : Import data sets
Select "Data" menu, then "Import", "From text file..."
Choose a local file on your disk containing datas you want to study.
Configure in import wizard, variables names, first line containing data, separators and correct the automatic assignment of variables to Symbolic or Numeric , clicking on it, if needed .
Process pretreatments.

Second step : Create corpus
In "Data" menu, click on "Export", "To corpus..." and choose the treatment you want to process.
All patterns are selected to learn by default.
Select validation and test patterns in data grid and right click on them if needed.
Select "Test list" to set those patterns as test ones.
You can save your corpus by clicking on "Corpus" menu and then "Save as...". (Click on "Load" to load a corpus.)

Third step : Create your data-mining model and enjoy
You can use corpus to create a neural network, a decision tree or a meta-learner by choosing "Network", "Tree" or "Meta-learner" menu.
Network are displayed in "Network" tab. You can see network topology and the values of each node while learning.
Unsupervised networks come with their "Input space" tab which show a 2D representation of your data. Use it to survey learning.
Networks are provided with their command panel (on bottom). Click on "Learn" to make the network learn. Error graph is display at right. Click on stop button to stop learning or change parameters.