Reference Manual: Introduction

From DSL

SMILEARN divides the process of learning graphical models into several conceptual steps. In the most general case, learning consists of the following three steps: (1) obtaining raw data from a data source, (2) preprocessing the data, and (3) applying a learning procedure that results in a model (or some equivalent structure). A raw data in Step 1 can be a text file, an external database or a set of observations associated with a model. The preprocessing step is an optional step that amounts to removing or filling missing values in the data set, discretization of continuous values, removing outliers, etc. Step 3 applies a learning algorithm to the preprocessed data. Which steps to perform depends on the problem at hand and on user needs.