cbcTools (development version)

cbcTools 0.7.1

Auto-encoding to dummy coding occurs for cbc_choices and cbc_power if the design is detected to have a no_choice option.
Updated some of the print methods so that they don’t error if the choices or design objects are modified by dplyr functions.

Re-configures the design encoding to use “standard” coding by default.
Reasoning for standard coding by default is for easier interpretation of summary metrics like balance and overlap.
New function cbc_encode() used to convert designs to dummy or effects coding.

Adds balance_by argument to force balanced sampling in designs with restricted or otherwise unbalanced levels across attributes.
Fixes issue where remove_dominant was not working if there was a no_choice option.

Improve the greedy methods to include proper handling of the dominance checking and overall efficiency improvements.

Added include_probs argument to cbc_design(), which includes predicted choice probabilities in the returned design data frame if include_probs = TRUE. Defaults to FALSE.

Major overhaul of the package with breaking changes.
New function, `cbc_priors()``. This allows users to specify a set of priors according to a wide variety of model specifications, including random parameters (with or without correlated heterogeneity), interactions, and “no choice” options. These priors can then be used to create designs and simulate choices.
Coefficients for levels of an attribute in cbc_priors() can be named vectors, addressing #24.
Major overhaul of the cbc_design() function, with entirely new algorithms for searching for designs
- One is “random”, three are frequency-based (“greedy”) algorithms, and three more are d-error minimizing algorithms.
- Old methods removed: "full", "orthogonal", "dopt", "CEA", and "Modfed"
- Bayesian D-efficient designs are now created based on the priors provided. With random parameters in the priors, a Bayesian D-efficient design will be created.
- New support for removing dominant alternatives from designs.
- New support for randomizing the order of questions and alternatives across respondents, addresses #29.
- New cbc_inspect() function for comprehensively inspecting designs.
- New cbc_compare() function for comparing designs.
- New functionality in cbc_power() for computing visualizing power analyses.

Further revisions to the method argument in the cbc_design() function.
Added the "random" and "dopt" methods.
Added restrictions so that orthogonal designs cannot use the label argument or restricted profile sets (as either of these would result in a non-orthogonal design).

Adjustments made to the method argument in the cbc_design() function in preparation for potentially adding new design methods.
Added the "orthogonal" option for generating orthogonal designs.

Another small bug fix in cbc_design() related to #16 where factor level ordering for categorical variables were being mis-ordered.
Updated how the method argument is handled by default in cbc_design() to be more flexible (anticipating other methods in the future).
Added keep_db_error arg to cbc_design().

Bug fix in cbc_design() where factor level ordering for categorical variables were being mis-ordered.
Added additional input check for appropriate priors in cbc_design().

Modify how restrictions are defined in the cbc_restrict() function to allow users to provide expressions.

Add cbc_restrict() function to improve UI for adding restrictions to profiles.
Remove previous approach to including restrictions in cbc_profiles().
Add new test cases

Bug fix: modify code in cbc_design() to avoid duplicate choice sets for the same respondents; addresses #7.
Bug fix: modify code in cbc_design() to allow Bayesian D-efficient designs with restricted profile sets; addresses #10 and #9.

Added initial integration with {idefix} packages for Bayesian D-efficient designs

Modified the argument of cbc_profiles() to ... so that the user no longer needs to create a separate list to define the attributes and levels.
Modified the arguments for the randN() and randLN() functions to mean and sd.
Improved printing of counts in cbc_balance() and cbc_overlap().
Updated names of random parameter models to match that of future logitr v0.6.0.
Updated documentation and examples for all functions.
Adding piping example to readme.