Skip to content

Instantly share code, notes, and snippets.

@mustmodify

mustmodify/gist:d04da7b2e7a953f152d1ab07514e4d85

Forked from anonymous/The score achieved from category based approach is in individual basis. WNFE obtained 0.599 from 2000 features, which is better than the PNTFA approach but not much satisfactory level as ICBA approach. PNTFA obtained least f-measure value of 0.475 fr
Last active June 13, 2016 21:51
Show Gist options
  • Save mustmodify/d04da7b2e7a953f152d1ab07514e4d85 to your computer and use it in GitHub Desktop.
Save mustmodify/d04da7b2e7a953f152d1ab07514e4d85 to your computer and use it in GitHub Desktop.
Download ZIP
Raw
gistfile1.txt
THe category-based scores are independent of the other. WNFE obtained 0.599 from 2000
features; this value makes it better than the PNTFA approach but not as good as ICBA approach.
PNTFA obtained the lowest f-measure value (0.475) from 500 features and the highest f-measure
value (0.515) using 9052 features. However, when statistical analysis was performed, there was
no difference between the result of 500 features and 9052 features. Thus, the f-measure value 0.475
(obtained using 500 features) was used. Merging best feature approach obtained an f-measure value of 0.661
from 2557 features. It was not as precise a score as is achieved by the collective-based approach.
The overall result of the multi feature experiment is in table 2. The collective-based approach incorporates
the collection of all kinds of features. The highest f-measure value achieved using that method was 0.705, using 20000 features.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment