search

prasanthmurali / bully-buster

A Naive Bayes Based Classifier to gauge instances of Cyber Bullying on Twitter. Initially developed at Boston Hacks 2017 and further worked upon by adding Topic Modelling (using Latent Dirichlet Allocation) to it.
1 stars 0 forks source link

first lda then nb #1

Open un-lock-me opened 7 years ago

un-lock-me commented 7 years ago

After searching a lot I found your code, before looking at your code I was thinking of applying LDA them NB on the LDA, But here I realized that you did the same but in another direction. I mean first NB then LDA.

May I ask you to explain more details why you firstly did the nb instead of LDA,

Also, which differences we might see if go in another direction(LDA then NB) according to your expriences,

I much appreciate your time,

prasvrin commented 7 years ago

Hi Saria, I wanted to perform a comparative analysis between Naive bayes and the subsequent performance increase using LDA. That's why you find me doing the other way !