def _extractFeatures(self):
goodWordFrequencies = {
word: TextHelpers.countwordoccurences(word, self._goodText) /
self._goodWordCount
for word in {x
for x in TextHelpers.getwords(self._goodText)}
}
badWordFrequencies = {
word: TextHelpers.countwordoccurences(word, self._badText) /
self._badWordCount
for word in {x
for x in TextHelpers.getwords(self._goodText)}
}
keptWords = sorted(goodWordFrequencies.keys(),
key=(lambda k: goodWordFrequencies[k] -
badWordFrequencies.get(k, 0)),
reverse=True)[:self._featureNumber]
result = [WordClassificationFeature(x) for x in keptWords]
return result
def _extractFeatures(self):
goodWords = {word for word in TextHelpers.getwords(self._goodText)}
shortGoodWords = {
word
for word in goodWords if len(word) <= self._minlength
}
shortWordGains = {x: self._stringGain(x) for x in shortGoodWords}
longGoodWords = set.difference(goodWords, shortGoodWords)
stringGains = self._getSubstringGains(longGoodWords)
allGains = stringGains
allGains.update(shortWordGains)
#allGains = { **stringGains, **shortWordGains }
keptWords = sorted(allGains.keys(),
key=(lambda k: allGains[k]),
reverse=True)[:self._featureNumber]
result = [
StringClassificationFeature(x) for x in keptWords
if allGains[x] > 1.0
]
return result