Shakespeare Statistics

Authorship of Fedele and Fortunio

BRD-Flagge auf deutsch All data were generated with R Stylo. See: ).

If several independent methods arrive at similar or identical results, it can be assumed that a theory has been consolidated. Classifications have proven to be reliable discrimination methods in the past. The classifiers nsc (nearest shrunken centroid), svm (support vector machine) and delta (according to Burrows) are equally suitable when a decision has to be made between two author candidates. The evaluations of the classifiers nsc, svm and delta with window sizes between 1000 and 8000 words at intervals of 1000 words were summarised in the following table for each 250-word segment. The vocabulary (mf1w) in columns B to I, as well as letter bi- and trigrams in columns J to Q and R to Y were available as variables (mf2c, mf3c). The first measurement of the 1000 window is recorded in its centre at 500 words (B2), the first measurement of the 8000 window analogously at 4000 words (I16). The same scheme applies to mf2c and mf3c.








Please find here the statistics of the attribution table.
Compare this evaluation with the results of Rolling Delta.