Text Analytics For Medical Device Innovation
- A consulting client supported a medical device manufacturer to develop a new product
- The client's innovation team wanted to learn from the mistakes of similar products and thereby create a product that met user expectations
- To collect and analyze competitor product reviews from Amazon.com to understand features that customers like vs. dislike, product shortcomings, etc.
- Designed a python web scrape to export the latest 5,000 reviews from Amazon.com, gathering data for the period Q1 2020 - Q2 2023
- Parsed and cleaned data in python for NLP and text analytics, leveraging methods that removed unnecessary punctuations, expanded contractions, lemmatized words to their root form, and removed stop words (frequently occurring words that do not add value to our analysis)
- Applied VADER and BERT for sentiment analysis and used Latent Dirichlet Allocation to discern dominant topics and their related sentiments in reviews
- Developed an ML model to pinpoint keywords driving 5-star reviews
- Tableau dashboard to navigate the results in a dynamic manner
- Key learnings and implications for the innovation team on:
- – Build quality ,Performance, Long-term usage, Value for money perception, Customer service, etc
