Training data set: 31,961 tweets

Test data set: 17,197 tweets

Text Procssing

• Coversion the text to lower case
• Remove the URL's
• Remove the special charactors
• Tokenization
• Lemmatization
• Vectorization

Models

1. Logistic Regresion model

• Accuracy: 95%
• Precision: 90%
• Recall: 31%

2. XGBoost model

• Accuracy: 94%
• Precision: 86%
• Recall: 16%

---

These are our findings:

---

Used:
Github : https://github.com/lumindak/sentiment_analysis