Optimal Subset Cardinality Regression (OSCAR) Models Using the L0-Pseudonorm
Return total cost of model fit based on provided kit/variable costs ve...
Return named vector of feature indices with a given k that are non-zer...
Showing oscar-objects
Return named vector of indices for kits with a given k that are non-ze...
S4-class for oscar
Extract coefficients of oscar-objects
oscar: Optimal Subset Cardinality Regression
Binary logical indicator matrix representation of an oscar object's co...
Visualize binary indicator matrix optionally coupled with cross-valida...
Bootstrap visualization with boxplot, percentage of new additions
Reformatting bootstrap output for cardinality k rows
Bootstrap heatmap plot for oscar models
Bootstrapping for oscar-fitted model objects
Visualize bootstrapping of a fit oscar object
Control OSCAR optimizer parameters
Return total cost of model fits if the cost is not included in the osc...
Cross-validation for oscar-fitted model objects over k-range
Visualize cross-validation as a function of k
Retrieve a set of pareto-optimal points for an oscar-model based on mo...
Visualize oscar model pareto front
Main OSCAR fitting function
Create a sparse matrix representation of betas as a function of k
Target function value and total kit cost as a function of number of ki...
Optimal Subset Cardinality Regression (OSCAR) models offer regularized linear regression using the L0-pseudonorm, conventionally known as the number of non-zero coefficients. The package estimates an optimal subset of features using the L0-penalization via cross-validation, bootstrapping and visual diagnostics. Effective Fortran implementations are offered along the package for finding optima for the DC-decomposition, which is used for transforming the discrete L0-regularized optimization problem into a continuous non-convex optimization task. These optimization modules include DBDC ('Double Bundle method for nonsmooth DC optimization' as described in Joki et al. (2018) <doi:10.1137/16M1115733>) and LMBM ('Limited Memory Bundle Method for large-scale nonsmooth optimization' as in Haarala et al. (2004) <doi:10.1080/10556780410001689225>). The OSCAR models are comprehensively exemplified in Halkola et al. (2023) <doi:10.1371/journal.pcbi.1010333>). Multiple regression model families are supported: Cox, logistic, and Gaussian.