3 modules to realize all your Machine Learning projects,
From the data preparation to the supervised classification
This preprocessing module improves the data preparation step, before using regular Machine Learning algorithms. In particular, it allows any classifier to be much more robust thanks to the recoding of data. Furthermore, the training stage is accelerated thanks to a reliable and efficient variable selection algorithm.
This module allows one to automatically produce Machine Learning models. The duration of the projects is considerably shortened and the required hardware resources are drastically reduced. The learned models are extremely reliable and its have an accuracy close to the optimal. The use of models in production is durable and secure.
This module automatically processes tabular data, as well as unstructured data of the kind "sequences" (eg texts, web sessions, application logs ...). The sequential rules are extracted from raw data, without any pre-processing. These rules describe the data and are used to learn an accurate and very robust model.
|Learning of univariate discretization and grouping models|
|Data recoding (gain of robustness)|
|Detection of non-informative variables (not correlated to the target)|
|Calibration of any probabilistic classifier|
|Automated learning of ensemble classifiers|
|Evaluation of the models|
|Predictions by using the models|
|Evaluation of drift levels|
|Seqeuntial rules extraction|
|Data recoding as binary variables|
|Automated learning of the models from sequential variables|
|These modules run as a command line (compatibility with all programing languages ) or it runs through a Python wrapper|