4 ms·
Show HN: Naive Bayes classifier for text categorization in five steps
- ColinWright 8y agoFrom the article: For an English spam classifier that considers all the words in the English language, the number of the words (n) is approximately 171,476. That's a remarkably precise number to be preceded by the word "approximately".
- gchavez2 8y agoAgree, that was odd, it now reads: "the number of the words (n) is approximately 170k" Thank you for the remark.
- jgrahamc 8y agoThis is not a bad explanation but when doing this practically it can be useful to take log() of the probabilities so that you work with sums of logs rather than multiplying small floats. http://getpopfile.org/docs/faq:bayesandlogs http://getpopfile.org/docs/faq:bayesandlogs
- gchavez2 8y agoThank you for the insight John, I have included your remark on the article.
- atum47 8y agoNice article, very glad to read it. Keep up the good work.
- gchavez2 8y agoThank you Victor, I enjoyed your JS articles too!