In this project, I have come up with novel approaches to make sense of Blenheim Palace’s customer survey data, with the help of different Natural Language Processing (NLP) and machine learning techniques. In addition, an extensive exploratory analysis revealed key insights and answers to set business questions. The data with a format of having multiple quantitative and free text responses were analyzed using combinations of techniques like clustering, sentiment analysis, topic modelling and semantic analysis to bring out very specific information, which could help Blenheim understand their customer experience better.
Blenheim Palace is a country house in Woodstock, Oxfordshire, England. It is the seat of the Dukes of Marlborough and the only non-royal, non-episcopal country house in England to hold the title of palace.
Some of the topics touched in this project are:
- Latent Dirichlet Allocation (LDA)
- Topic modelling (BERTopic) - both manual and library implementations
- Sentiment Analysis (Lexicon based and pre-trained models)
- Semantic search and analysis
- Clustering (k-Means algorithm)
- Wordclouds
- Supervised ML models
- Data Visualization
Please note that I can't put the data here due to confidentiality. You're most welcome to use your own survey data, as the code written here in this project can be generalized for other free text responses.