SIPINA OVERVIEW

SIPINA is especially intended to decision trees induction (says also Classification Trees). Use TANAGRA if you want other kind of analyses such as clustering algorithms, svm, factorial analysis, statistical hypothesis testing, etc.

SIPINA is a Data Mining Software which implements various supervised learning paradigms. SIPINA is an academic tool, it is free for all kind of activities.

SIPINA is distributed on the web since 1995, it runs under Windows OS (W95 and later) . My main preoccupation is to build an experimentation engine for research activities. I use Sipina for my researches publications (see publications) and, on sharing Sipina on the web, I hope it will help other researchers.

SIPINA is mainly a Classification Tree Software (specialized on Classification Trees algorithms such as ID3, CHAID, C4.5, ASSISTANT-86, etc....). But, other supervised methods are also available (e.g. k-NN, Multilayer perceptron, Naive Bayes, etc.). We can perform perfomances comparison and model selection.

SIPINA can be freely downloaded here.


Fig. Running SIPINA on the famous IRIS dataset

SIPINA theoretical limitation are 16,384 attributes and 500,000,000 examples. Because it loads all dataset in memory before the learning process, the true limitation is the memory that it can address (3 GB under 32-bit Windows, 4 GB under 64-bit Windows). SIPINA can handle both continuous and discrete attributes. The number of values of a discrete attribute is limited to 16,384.

Ricco Rakotomalala
Lyon, FRANCE.