OCHEM was designed as a user-friendly web tool for model development.
It provides an access to about 10 different machine learning methods and more than 20 different descriptors.
OCHEM can easily develop models with >300k molecules and has handled data matrices with up to 2*1011 entries (~300k molecules X 700,000 descriptors). For analysis of ToxCast data more than 20k models were developed as analysed elsewhere.
The following tutorial explains the development of models using OCHEM.