Usage

uv sync
source .venv/bin/activate
jupyter-lab

The report is captured in project.ipynb.

AI tool usage:

• chatgpt and claude desktop

Future work

• Caption analysis to map captions to certain features of captions themes and check if there is any high performing cluster
• Caption feature could include aspects like caption length, whether it has CTAs or not etc.
• Better NLP tools like LLMs to better understand the sentiments of the comments instead of textblob, which is a bit limited in its capabilities.
• A key missing information is the analysis so far is the content of the post. It is very likely that the actual content of the post plays a dominant role in driving engagement. The information themes can be extracted through image/video analysis of post content to understand if there are common themes that drive engagement.