Quick Start for Practitioner

Usually, practitioners are likely to use fully implemented end-to-end methods. We have designed skscope to support powerful SCO methods for non-trivially and practically important problems to accommodate this need. skscope serves as a machine learning factory continuously producing easy-to-use end-to-end sparse learning algorithms.

The software architecture of skscope.

Specifically, the submodule skscope.skmodel in skscope includes the implementation of practically valued SCO methods that can be directly used by practitioners. More importantly, these methods are designed to be compatible with the sklearn library. This allows Python users to use a familiar sklearn API to train models and easily create sklearn pipelines incorporating these models. Currently, these end-to-end methods supported by skscope are summarized in this table:

Some application-oriented interfaces implemented in the module skscope.skmodel in skscope.

skmodel	Description	Document
PortfolioSelection	Construct sparse Markowitz portfolio	Portfolio selection
NonlinearSelection	Select relevant features with nonlinear effect	Non-linear feature selection via HSIC-SCOPE
RobustRegression	A robust regression dealing with outliers	Robust regression
MultivariateFailure	Multivariate failure time model in survival analysis	Multivariate failure time model
IsotonicRegression	Fit the data with a non-decreasing curve	Isotonic Regression