NEWS (original) (raw)

misaem 1.1.0

A release improving stability and speed:

Vectorization of the function miss.glm.fit
Improve the stability of miss.glm.fit by catching infinity ratio
Catch some convergence issues of miss.glm.fit and raise appropriate error
Allow the use of regularization in the logistic regression, by using glmnet package

A minor release mainly fixing bugs and typos and also serving for authorized maintainer takeover:

Fix a bug in miss.lm.fit, the definition of p, the number of covariates, is now moved to the beginning of the function. A
Fix another bug in miss.lm.fit, replacing is(x, “data.frame”) with !is(x, “matrix”) to make the parameter consistency check also pass for a single observation in x.
Fix some typos in the descriptions of the functions for linear regression model.
Move mice package from imported to suggested packages.
Takeover of maintenance from Wei Jiang by Julie Josse, who was co-author of the package from the beginning.

A major release mainly update the follows:

Add linear regression with missing values, see the functionmiss.lm.
The main function for logistic regression is changed tomiss.glm.
The functions mimics the structure of widely used functionslm and glm for the case without missing values.
Update misaem tutorial.
Update README.md.
Typos fixed.

A minor release mainly fixing bug:

Replace if (class(obj) == "data.frame") { ..... } withif (is(obj, "data.frame")) { ..... }. As within the formal S4 class system, a matrix obj could well be of a class that extends matrix, such as “matrix” “array” (length = 2).

A minor release mainly fixing bugs and typos:

Fix a bug in model_selection, now it can correctly perform model selection if the full model is the best model.
In pred_saem, two methods for prediction of test set with missingness are provided.
Fix some typos in the vignettes. The length of codes now fits the page wide of html.
Delete unused Imports magrittr in DESCRIPTION file.
Change the index of vignettes from ‘SAEM’ to ‘misaem tutorial’.
Update README.md.