I am working on Sentiment Analysis and have a dataset that has 48000 tweets with labels. 7000 Negative, 17000 Neutral and 24000 Positive Tweets. Is my dataset is balanced and if not how can i balanced it?
Balancing is not necessary, however., if you are still interested in handling balancing issues, I suggest you to follow https://towardsdatascience.com/handling-imbalanced-datasets-in-machine-learning-7a0e84220f28
https://www.analyticsvidhya.com/blog/2017/03/imbalanced-data-classification/
On your DS the amount of positive is 3.4 times more, that amount of negative - so it is evident, that DS is imbalanced. To make balanced Ds it is possible to use different techniques - random under-sampling (RUS), random over-sampling (ROS), SMOTE, etc.
To add to Sergey Porotsky 's answer - you could also just collect more data.
There's lot of public datasets concerning sentiment analysis of tweets. Here's a link to a search in Google's DataSetSearch:
https://datasetsearch.research.google.com/search?query=twitter%20sentiment&docid=NiCF0tlDI1nnialTAAAAAA%3D%3D
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics