Description Usage Arguments Details Value See Also
This function will calculate the Weight of Evidence for each bin of a given variable, according to a specified binary response. A vector of bins can be supplied, or a pre-binned variable can be supplied.
1 | woe_single(dat, var, bins = 0, response, warn = TRUE, auto_merge = FALSE)
|
dat |
The dataset containing the response and desired variable |
var |
Character string containing the name of the variable |
bins |
Either a numeric vector specifying the cutpoints for the desired variable, or 0. If 0, the function assumes that the variable is already binned, and throws an error if it is not of type "factor" |
response |
Character string specifying the name of the response variable |
warn |
Logical- should the function print a warning if a given bin contains only one level of the response (and hence has a WOE of +/- Inf)? Recommended to keep this to TRUE; FALSE option is provided for using this function in programming |
auto_merge |
Logical. If bins contain only one value of the response, should they be merged with an adjacent bin? If set to TRUE and warn is set to FALSE, the function can modify your data without telling you- be careful |
woe_single
allows the user to easily calculate Weights of
Evidence for arbitrary bins for a given variable.
The function will throw an error if bins
is 0 and the
supplied variable is not a factor. The idea is for the user to
supply their own bins obtained using, for example, one of the useful
functions in the modellingTools
package, such as
vector_bin
. When bins have only one level of the
response, the WOE will be +/-; by default the function will not prevent this
and issue a warning. You can turn off warnings and have the function auto-merge
bins using the supplied options, but be sure to inspect the results carefully.
A two column tbl
containing the bin and the WOE value
information_value
Other metrics: information_value
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.