In mdlincoln/kwm: Explicit Regex Matching Implemented As Model-like Objects

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)

kwm

kwm provides very simiple wrapper functions to produce KeyWord Models that produce classification predictions based on explicit lists of regular expression pattern matches. By supplying a generic prediction function for such lists, it is easy to compare the performance of very simple regex matching to other, more complicated text classification models within the same pipeline.

Installation

You can install kwm from github with:

# install.packages("devtools")
devtools::install_github("mdlincoln/kwm")

Example

library(kwm)

month_df <- data.frame(month = month.name, stringsAsFactors = FALSE)

# Locate all matches that INCLUDE either "a" or "e" but EXCLUDE any ending in "r"
month_model <- kwm(include = c("a", "e"), exclude = "r$", varname = "month")

predict(month_model, newdata = month_df, return_names = TRUE)

# You can pass options to the underlying search function as well
caseless_month_model <- kwm(include = c("a", "e"), exclude = "r$", 
                            varname = "month", 
                            search_opts = list(ignore_case = TRUE))

predict(caseless_month_model, newdata = month_df, return_names = TRUE)