Currier A/B split is not what we thought it was!

[Labyrinthinesecurity]

I have fixed the <-> parsing problem and rerun the tests: in the Herbal section, which is the most critical for extracting the switch feature, the impact is not statistically significant.

Before:
Herbal folios: 125
    Classifiable (>= 5 words): 124
    sigma=1 (cho-dom): 92
    sigma=0 (che-dom): 32
    Mean r_cho (sigma=1): 0.752 +/- 0.134
    Mean r_cho (sigma=0): 0.170 +/- 0.095

After:
Herbal folios: 125
    Classifiable (>= 5 words): 124
    sigma=1 (cho-dom): 92
    sigma=0 (che-dom): 32
    Mean r_cho (sigma=1): 0.752 +/- 0.134
    Mean r_cho (sigma=0): 0.169 +/- 0.087

In You are not allowed to view links. Register or Login to view., you wanna look at the last table called "APPENDIX: HERBAL FOLIO DETAILS" for per folio computations in Herbal section, here are the first few lines (sig 1 means Currier A, sig 0 means Currier B):

==============================================================================
APPENDIX: HERBAL FOLIO DETAIL
==============================================================================

  Folio    sig n_cho n_che  r_cho  Conf  e/ch    d/l Words
  ----------------------------------------------------------
  f1v        1    27    6  0.818  1.000  0.306  0.478    85
  f2r        1    18    6  0.750  1.000  0.372  0.559  100
  f2v        1    22    5  0.815  1.000  0.321  0.556    56
  f3r        1    30    17  0.638  1.000  0.443  0.364  111
  f3v        1    22    6  0.786  1.000  0.340  0.419    83
  f4r        1    16    3  0.842  1.000  0.258  0.500    63
  f4v        1    27    11  0.711  1.000  0.439  0.677    78
  f5r        1    14    7  0.667  1.000  0.545  0.750    53
  f5v        1    17    4  0.810  1.000  0.269  0.556    43
...


Interestingly, the switchable templates have now more impact on the Currier classification the switch now accounts for 9% of the variance (versus 8% before), and we have 4 more templates, 2 of which are new switchable templates.

chXckhy                0.909  0.237    11    59  0.672 S-switchable
shXcthy                0.900  0.182    10    22  0.718 S-switchable