NB: Due to the nature of this project, code cannot be shared publicly.
The main aim of this project was to classify people's sentiment towards climate change using data from X formerly known as Twitter, classifying tweets as either being positive, neutral or negative.
At least 85 per cent of the global population has experienced weather events made worse by climate change. Understanding people's sentiments on this topic is crucial in today's environmentally conscious world, and for businesses to remain relevant in today's market, they need to be conscious of the latest trends and sentiments of their main client base towards social issues. This can be achieved by analyzing individuals' tweets to determine their opinions on climate change.
- Analyse and identify key insights in the Twitter dataset (Jupyter Notebook)
- Create a sentiment analysis classification model (Jupyter Notebook)
- Create a user-friendly sentiment analysis app (Streamlit)
- Report findings (PowerPoint presentation)
- Python (Jupyter Notebook, Streamlit (VScode), scikit-learn, nltk, imblearn)
- Comet
- Github
- AWS EC2
The following is a sample of the analysis and insights drawn while working on the project
- StephenSchlegel 507
- SenSanders 387
- BernieSanders 194
- NatGeoChannel 161
- thehill 138
- CNN 132
- SethMacFarlane 126
- ClimateCentral 109
- climatehawk1 106
- nytimes 99
Through our exploration of the Twitter dataset, we found a higher number of retweets over original tweets on the subject matter. With this information, we decided to find out who were the most retweeted accounts and we identified the popular accounts whose views on climate change were shared. Stephen Schlegel and Bernie Sanders were the most retweeted accounts, an indication that they have a strong following and their views on climate change invoke some action from their followers.
A word cloud on the most used hashtags associated with climate change.
0228.1.mp4
Mantsali Sekoli - @Mantsali
Tercius Mapholo - @TerciusMapholo
Fumani Thibela - @Fumani09
Colette Muiruri Wamuchie - @muiruric
Link to Kaggle competition - https://www.kaggle.com/competitions/edsa-sentiment-classification