Set a maximum value to the frequency of pairs

Transforms the frequency in a co-occurrence dataframe, setting an stipulated maximum value of words frequency per document. The input must be the tibble from cooccur_words(output=df2). It groups the data by two columns, sums the counts within each group, and arranges the results in descending order. It is useful to reduce the maximum value that each document contribute to the corpus with its node pairs, avoiding that a highly frequent word pairs in a single document or in only a few documents, but rare or absent in other documents be taken as representative of the whole corpus.

Usage

reduce_freq(cooc, threshold)

Arguments

cooc

A dataframe from cooccur_words(output=df2).

threshold

The threshold max value above which counts are capped in each document.

Value

A transformed dataframe with summed counts arranged by their magnitude.

Examples

# Example usage:
cooc_data <- data.frame(n1 = c(1, 1, 2), n2 = c(3, 4, 3), n = c(5, 6, 7))
cooc_data
#>   n1 n2 n
#> 1  1  3 5
#> 2  1  4 6
#> 3  2  3 7
reduce_freq(cooc_data, threshold = 5)
#> # A tibble: 3 × 3
#>      n1    n2     n
#>   <dbl> <dbl> <dbl>
#> 1     1     3     5
#> 2     1     4     5
#> 3     2     3     5