This project aims to create a data analytics tool that can be used by business owners to understand their customers. Customer reviews can be used to create a better impact on businesses. This tool demonstrates the kind of analysis that can be performed on customer reviews. I am focussing on 101 businesses due to hardware limitations, but this can be very easily scaled to process much higher volume.
- Using NLP libraries like VaderSentiment, BERT Transformers customer reviews are boiled down to important entities, and the sentiment of each entity is captured.
- Entities represent the specific aspects that customers discuss in their reviews, such as food, order, service, time, staff, and ambiance.
- These entities will be clustered into similar buckets. (balance will have to be maintained between too many buckets - too specific, or too few - loss of information.
- Footfall analysis, comparing the negative effect of one entity over another on star rating, performance attribution, and many more kinds of analysis will be performed and presented in a drill-downable dashboard.
- Report generation and broadcasting it to relevant personnel will also be possible.
- Added functionality to visualize average rating in various levels of analysis. Yearly, monthly, daily, hourly.
Steps and commands to setup the environment for this project
conda create -n capstone_env python=3.12 conda activate capstone_env
conda install ipykernel python -m ipykernel install --user --name=capstone_env jupyter notebook
create a new jupyter notebook by selecting "capstone_env" as the kernel