Load TrainSet: 1181/3000 positive/negative samples.
Vocabulary: 6666 items.
Vocabulary (stem): 5106 items.
Load TestSet: 1182/3000 positive/negative samples.
noStem | Stem | |
---|---|---|
freq | 89.43% | 90.20% |
bin | 89.41% | 89.93% |
tfidf | 88.47% | 87.97% |
parameters: learning rate = 0.10, iternum = 30000
Without regularization:
noStem | Stem | |
---|---|---|
freq | 90.00% | 89.89% |
bin | 89.69% | 90.03% |
tfidf | 90.17% | 90.15% |
With L2 regularization:
noStem | Stem | |
---|---|---|
freq | 90.05% | 90.27% |
bin | 89.50% | 89.91% |
tfidf | 90.17% | 89.43% |
Thanks