I am working on sentiment analysis on twitter data. I have tried with couple of Naive Bayes models like Baseline Naive Bayes, Multinomial NB, Bernoulli NB, Semi-supervised NB. My question here is to understand if there is a way we can combine the models in below two way to improve perforrmance?
Thanks
Naive way: 1) Take the average the scores and do final prediction. 2) Do predictions and then take maximum vote
Machine Learned way: You need to use ensemble approach - Do stacking. Prepare a meta-model over these models.
You can refer this https://machinelearningmastery.com/machine-learning-ensembles-with-r/ . Things are done here using caret package in R.
Similar things you can do it in python also using flagship library sklearn .
