Hi
I have used Multinomial naive bayes for this challenge with CountVectorizer ngram_range=(1,3), after submitting predictions I got 88% accuracy with this. So what shall I do to increase accuracy, shall I increase ngram_range ?
Query regarding accuracy
Hey @pasta
Accuracy always depends on few things.
- Data Preprocessing
- Features Selected
- Model Building
etc etc
If you use naive bayes 86-87 is the max accuracy you will be able to get as far as i know (there is always scope for improvement)
I used NeuralNets and got 90% accuracy, and with proper hyper parameter tuning in NeuralNets, hopefully you will be able to achieve around 92.
So, you can try some hyperparamer tuning or a different model !
I hope this helps you clear the doubt !
Happy Learning 