Regression (regression)

Orange implements a set of methods for regression modeling, that is, where the outcome - dependent variable is real-valued:

• Linear regression (linear)
• Lasso regression (lasso)
• Partial least sqaures regression (PLS)
• Regression trees (tree)
• Mean (mean)

Notice that the dependent variable is in this documentation and in the implementation referred to as class variable. See also the documentation on Classification (classification) for information on how to fit models and use them for prediction.

Base class for regression

All regression learners are inherited from BaseRegressionLearner.

class Orange.regression.base.BaseRegressionLearner(imputer=None, continuizer=None)

Fitting regressors typically requires data that has only continuous-valued features and no missing values. This class provides methods for appropriate transformation of the data and serves as a base class for most regressor classes.

continuize_table(table)

Replace discrete variables with continuous and return a new instance of Orange.data.Table.

Parameters:table (Orange.data.Table) – data instances.

impute_table(table)

Impute missing values and return a new Orange.data.Table object

Parameters:table (Orange.data.Table) – data instances.

set_continuizer(continuizer=None)

Set the continuizer of the discrete variables

Parameters:continuizer (None or Orange.data.continuization.DomainContinuizer) – function which replaces the categorical (dicrete) variables with numerical variables. If None, the default continuizer is used

set_imputer(imputer=None)

Set the imputer for missing data.

Parameters:imputer (None or Orange.feature.imputation.ModelConstructor) – function which constructs the imputer for the missing values, if None, the default imputer replaces missing continuous data with the average of the corresponding variable and missing discrete data with the most frequent value.