can you plz suggest something for more accuracy.
Sir i am getting only 75% of accuracy
hey @Jan19LPN0013 ,
can you please provide drive link to your code file , so that i can understand your approach and therefore can suggest you something to improve it.
sir here is my code.
hey @Jan19LPN0013,
Suggestions for current approach :
- Improve your text cleaning function as currently i can see there are many incomplete words in your cleaned corpus along with numeric data also. This kind of text decreases your model accuracy . So you can have over it.
Your code is really good and hence aren’t much changes that can be made to your code to improve accuracy,but if you try another approach then it can be surely improved.
Sentiment Analysis depends on how much better you can understand the semantic meaning of the words in text , how these words are placed in a sentence .
To understand text better we use word embeddings , a n-dimension vector which a word or a text in a vector space.
For now , you can have reference of this page to understand about word embeddings and how they can generated. Learn how to implement them in your code and increase your accuracy from 75% to 85-90%.
Else if you face any problem in these , you can surely ask me here or in chat .
so i have to remove numeric data and I have gone through that link you provided I did things written in that except n-grams so how would I include bigram in my code plz explain?
sir how to chat with you?
hey @Jan19LPN0013, for word embeddings you can use pretrained word embeddings like Glove , FastText , etc. Just search about it . They are very easy to understand.
In which way you want to chat?
as you suggest for chat.
you can message me on my whatsapp at 8630831390 ,or if you want to chat here at this platform then we can start conversation here itself.
ohk thankyou sir for providing your number
no problem , you can also call me on the same number if you get stucked somewhere.
Is your doubt resolved now?
yes thankyou sir for your help
