https://github.com/brayvid/tweet-sentiment-classifier

Flatiron School Data Science Bootcamp Phase 3 Project
https://github.com/brayvid/tweet-sentiment-classifier

classification data-science kaggle machine-learning scikit-learn sentiment-analysis twitter

Last synced: 5 days ago
JSON representation

Flatiron School Data Science Bootcamp Phase 3 Project

• Host: GitHub
• URL: https://github.com/brayvid/tweet-sentiment-classifier
• Owner: brayvid
• Created: 2024-07-09T23:07:56.000Z (6 months ago)
• Default Branch: main
• Last Pushed: 2024-07-17T22:45:38.000Z (6 months ago)
• Last Synced: 2024-11-12T12:18:54.770Z (2 months ago)
• Topics: classification, data-science, kaggle, machine-learning, scikit-learn, sentiment-analysis, twitter
• Language: Jupyter Notebook
• Homepage: https://colab.research.google.com/github/brayvid/tweet-sentiment-classifier/blob/main/tweet_sentiment_classifier.ipynb
• Size: 21.6 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Tweet Sentiment Classifier



Blake Rayvid - https://github.com/brayvid



Flatiron School Data Science Bootcamp Phase 3 Project

Presentation slides


## Business problem

Brand reputation management

Monitor brand perception by correctly classifying new tweets as positive, negative or neutral.



Analyze negative feedback for insights into product weaknesses and use this to drive improvements.



Identify accounts with consistent positive sentiment and offer to collaborate.



Time launches of new products during periods of high positive sentiment.





## Dataset

https://www.kaggle.com/datasets/yasserh/twitter-tweets-sentiment-dataset



Three classes: positive, negative, neutral in column called sentiment.



27,000 tweets formatted as strings in text column.





selected_text is an additional column containing the substring of each tweet relevant to classification.









## Results

I tried several model types, and a Support Vector Classifier (SVC) applied to ‘selected_text’ yielded the best performance. Test set results are summarized below, with precision and recall scores per class and a confusion matrix. Test accuracy was 83%.

Label

Precision

Recall

negative

83%

77%

neutral

78%

91%

positive

93%

80%



## Next steps



Try Word2Vec semantic embedding instead of frequency-based TF-IDF.



Investigate dimensionality reduction with UMAP or t-SNE.



Deploy to a web service to classify new tweets in real time.





This project highlights the importance of sentiment analysis in brand reputation management and provides a foundation for further development and deployment in a real-world setting.