Table 1 Dataset distribution for sentiment analysis.
From: A hybrid deep learning model for sentiment analysis of COVID-19 tweets with class balancing
Sentiment classes | Training set | Test set |
|---|---|---|
Binary classification (Negative, Positive) | ||
  Negative | 23,111 | 2252 |
  Positive | 18,046 | 1546 |
Three-class classification (Negative, Neutral, Positive) | ||
  Negative | 15,398 | 1633 |
  Neutral | 7713 | 619 |
  Positive | 18,046 | 1546 |
Five-class classification (Negative, Neutral, Positive, Extremely Negative, Extremely Positive) | ||
  Negative | 9917 | 1041 |
  Neutral | 7713 | 619 |
  Positive | 11,422 | 947 |
  Extremely negative | 5481 | 592 |
  Extremely positive | 6624 | 599 |
