A list containing the following elements, to be passed to the ReadMe
function. To call ReadMe with your own choice of values, edit this list
or pass changes separately as arguments.
- trainingset
- Binary table indicating which words, among those
words that satisfy threshold, appear in which text in the
training set. Words appearing in all or none of the texts are
omitted regardless of threshold.
- testset
- Same format as trainingset, but for test set of texts.
- formula
- Formula to use in the call to function VA, where the
main computation is done. Defaults to include all words that
statisfy threshold and appear less than 100% of the time as
dependent variables. ``truth'' from control file is the explanatory
variable.
- features
- Number of features to use in each subset in VA.
Corresponds to nsymp in VA. Defaults to 15. See VA or
ReadMe function documentation for details.
- n.subset
- Number of subsets to use in VA. Defaults to 300. See
VA or ReadMe function documentation for details. Larger numbers
produce more precision.
- prob.wt
- Vector of probability weights for the features to be
employed by VA. Must have length equal to the number of features in
the formula. Defaults to 1. See VA or ReadMe function documentation
for details.
- boot.se
- Use bootstraping in VA to compute standard errors?
Defaults to FALSE. Bootstrapping produces standard errors but is
time-intensive. See VA or ReadMe function documentation for
details.
- nboot
- How many bootstrapping samples for VA? Defaults to 300.
Ignored unless boot.se set to TRUE. See VA or ReadMe function
documentation for details.
- printit
- Print progress of VA function? Defaults to TRUE. See VA
or ReadMe function documentation for details.
Gary King
2011-07-12