NEWS

Qval 1.2.4 (2025-11-25)

Added - is.Qident function to iteratively explore the attribute hierarchy, along with its corresponding S3 methods.
Added - att.hierarchy function to test the identifiability of the Q-matrix, along with its corresponding S3 methods.
Added - sim.data function can randomly generate data with hierarchical structures.
Added - CDM function can perform parameter estimation that incorporates hierarchical structures.
Added - get.beta, get.PVAF, get.R2 and get.priority functions can perform parameter estimation that incorporates hierarchical structures.
Added - The validation function will be computed in parallel when the method is set to "Wlad", "beta", or "MLR-B".

Added - MLR-B accepts alpha.level.
Added - Simplified the output format of fit.
Added - The new fit class, which provides comprehensive S3 methods.
Change - Improved the documentation for etract.
Change - etract is compatible with the etract function in the GDINA package.
Change - Added a default value of i = 1 for the Hull plot.
Change - Enhanced the data type compatibility of get.beta, get.priority, get.PVAF, and get.R2.
Fixed - Resolved warning messages generated by the package.

Added - The summary method has been added for the CDM, simData, and validation classes.
Added - The plot method has been added for the CDM and simData classes.
Added The updata method has been added for the CDM, simData, and validation classes.
Change - Further standardized the hypothesis testing procedure for Wald.
Change - Improved the question names for Beta, Priority, PVAF, and R2 to be consistent with the original Q matrix.
Fixed - Resolved the issue occurring when method = 'beta'.
Fixed - Improved the code standards for S3 methods

Added - Separately provided beta metrics.
Added - Appropriate S3 methods have been provided for the CDM, simData and validation classes defined in the package to offer better interaction for users.
Change - Optimized the efficiency of all the code, making it more in line with the characteristics of the R language.
Change - The usage of the validation function has been changed when not using the iterative process.
Fixed - Resolved the issue occurring when method = 'Wald'.
Fixed - Resolved the issue occurring when method = 'beta'.
Fixed - Modified the manual description.

Added - A search algorithm for 'test.att' has been added for GDI, Hull, and beta.

Added - Added the URL of the online manual to the DESCRIPTION field.
Added - A new Q-matrix validation method, the beta (β) method, has been added.
Added - Wald now includes the SSA search method and the item.level iteration level.
Added - Added a PVAF cut-off point for GDI predicted by logistic regression (Najera et al., 2019).
Changed - Some of the important code is rewritten in C++ to accelerate the Q-matrix validation.
Fixed - Resolved the issue occurring when iter.level = 'item'.

Added - Provided a more comprehensive manual description for the validation function.
Added - Provided a more comprehensive manual description for the sim.Q function.
Added - Added a function get.Rmatrix for calculating the restriction matrix.
Added - Added a function get.priority for calculating the priority of attribute.
Added - Added a flexible function Wald.test for calculating the Wald test.
Added - Added SSA search for the Hull method.
Added - Added the plotting function plot.Hull for the Hull plot.
Changed - Rewrote the Wald method with the following updates: 1. If the search method is stepwise or forward, it will call the Qval function from the GDINA package. 2. If the search method is PAA, the search will follow the PAA. 3. The information matrix used is the full information matrix, implemented by the internal function inverse_crossprod from the GDINA package.
Changed - Renamed the functions getQRR, getVRR, getTPR, getTNR, getUSR, and getOSR to zQRR, zVRR, zTPR, zTNR, zUSR, and zOSR, respectively.