Twitter Sentiment Analysis
2020Collaborative
PythonBERTTensorFlowTransfer LearningNLPTwitter API
Sentiment analysis of meal kit services on Twitter using a pretrained BERT model — classifying tweets as positive, negative, or neutral per brand.
Started as a class project for Stat 576 (Data Informatics) at CSULB — my first hands-on experience with neural networks and transfer learning. The task: use Twitter's API to pull brand mentions for major meal kit services and classify each tweet's sentiment.
A pretrained BERT model (run via Google Colab with TensorFlow) beat simpler approaches easily. The results gave a real-time read on brand perception that complemented Pick a Kit's star-rating data.