Methods update

[mwelz]

• Added a class structure for the accessor functions
• Added new function get_best()
• Improved documentation

[aalfons]

Thanks, in general it looks very nice and the class structure is good, but there are some details that need attention.

Naming can be improved:

• It's JSS style to use "p value" instead of "p-value". Since we have to use that in the text of the paper, it will be nice to match the R output to the text and use "p value" as the column name in the printed output. Accordingly, the component in the "BLP_info" objects etc. should be called "p_value" to be consistent with the other components, where you use underscores to separate words (e.g., confidence_interval).
• In the class, the component is called confidence_interval, but the printed output uses CB for confidence bounds. Also, "CB lower" is a rather cumbersome abbreviation for "lower confidence bound" since the words are out of order. I'd suggest to use "Lower CB" to reflect the correct word order, "CI lower" to refer to the confidence interval lower bound (which would be more consistent with the naming of the component in the object), or simply "Lower". Also the printed information on the confidence level should change accordingly. For example, if you use the latter, the message should be something like Confidence level of confidence interval [Lower, Upper]:.

In the documentation, not all instances of class names are put in quotation marks, and not all function names are followed by parentheses:

• There are several instances on the help file of GenericML() where these are missing.
• On the help files of the accessor functions, you use: An object of class "ggplot" (see ggplot) and you link the second use of ggplot in parentheses. You can just use (removing the part in parentheses): An object of class \code{"\link[ggplot2]{ggplot}"}.
• Also check all other help files carefully.

These are details, but attention to those details matters if we want people to be confident about using the package.

[mwelz]

Thank you for your comments, @aalfons! I have implemented your suggested changes; I ultimately went for "CI lower etc. as printed name of the confidence bounds.