lisc.analysis.counts.compute_similarity¶
- lisc.analysis.counts.compute_similarity(data, dim='A')[source]¶
Calculate the similarity across the co-occurrence data.
- Parameters
- data2d array
Counts of co-occurrence of terms.
- dim{‘A’, ‘B’}, optional
Which set of terms to compute similarity across. ‘A’ is equivalent to across rows, ‘B’ to across columns.
- Returns
- cosine2d array
The cosine similarity of the co-occurrence data.
Notes
This function computes the cosine similarity.
Cosine similarity is normalized, so this function will give the same result if computed on raw counts, or normalized data.
The implementation is adapted from here: https://stackoverflow.com/a/20687984