v1.1.1

Bug fix release

Fixed a critical bug that prevented many of the B-methods from converging in small datasets.

v1.1

Includes new/improved learners

• highly adaptive lasso (hal, ahal, ahalb)
• lasso with cross validation (cvlasso)

Features

• Cross validation now respects clustering/individuals with multiple records via "id" macro variable - this allows for survival analysis via discrete hazard functions! The SuperLearner macro now implements every primary feature of the R super learner package.

Bug fixes

• major: hard coded outcome variable names in some learners
• minor: improved macro variable scoping, which may have caused some errors in limited circumstances

Aesthetics

• improved responsiveness of output

v1.0

Official stable release in preparation for manuscript

Beta 2

Multiple bug fixes including full implementation of seed values, fixed ridge regression, several R learners

Features: added Ridge regression to estimate super learner fit, harmonized adaptive random forest super learner with other methods to give identical output

Beta release

Pre-release of version 1.0