Calculate string distance(s) & group into networks

similar_strings.R

#### libraries ----

# install.packages("intergraph")

library(dplyr)
library(purrr)
library(furrr)
library(GGally)
library(igraph)
library(stringdist)
library(tidyr) # for expand_grid

#### environment parameters ----

plan(multiprocess)

#### helper function(s) ----

similar_strings <- function(df, x, y = NULL, dist = 0.7, ...) {
  
  x <- enquo(x)
  y <- enquo(y)
  
  if (quo_name(y) == "NULL")
    y <- x
  
  dat <- expand_grid(x = df[[quo_name(x)]],
                     y = df[[quo_name(y)]]) %>%
    mutate(string_dist = pmap_dbl(., ~stringsim(.x, .y))) %>%
    filter(string_dist < 1, string_dist >= dist)
  
  out <- list()
  
  out$distances <- dat
  
  out$graph <- dat %>%
    select(x, y) %>%
    graph_from_data_frame()

  out$networks <- decompose.graph(out$graph)
  
  out$similar_strings <- lapply(out$networks, function(x) V(x)$name)

  out
  
}

#### load data ----

data(mtcars)

dat_mtcars <- as_tibble(mtcars) %>%
  mutate(models = rownames(mtcars))

#### Calculate string distance(s) & group into networks ----

dat <- similar_strings(dat_mtcars, models, dist = 0.6)

# list of similar strings
dat$similar_strings

# plot full graph

ggnet2(dat$graph, label = TRUE, layout.exp = 1.1)

# plot networks individually

lapply(dat$networks, ggnet2, label = TRUE, layout.exp = 1.1)