Understanding Hawaii Tourist's Preference - NLP
- ckevinkusuma
- Jun 3, 2021
- 2 min read
Summary
This project will create a professional and valuable analysis from customers’ reviews on specific products/activities of Polynesian Cultural Center, a cultural theme park in Hawaii. While primarily designed to fulfill a required course work from MKTG 6640, this project’s secondary role is to be used by the management of the Polynesian Cultural Center as a credible reference for their pursuit of service excellence. Through its CMO/vice president, the company has agreed to grant access to a subset of its customer satisfaction survey data for this project. Polynesian Cultural Center is a non-profit organization that dedicates its efforts in preserving Polynesian cultures, including Hawaii, Tahiti, Samoa, Tonga, Aotearoa, and Fiji, while giving opportunities to underrepresented students from Asia-Pacific regions to pursue higher education at nearby Brigham Young University – Hawaii. The analysis resulted from this project will go beyond assisting the company to better its services. It will also support the preservation of Polynesian heritage, educate future generations, and provide opportunities to thousands of less-fortunate students.
This attraction company offers authentic Polynesian cultural experiences (which include Hawaiian culture) to visitors from every corner of the world, with one million visitors annually. They offer a wide range of activities and services to customers such as bus pick up, shows, cultural activities, retail stores, restaurants, concessions, and tours. To enhance customer experience and maintain product quality, the company collects satisfaction survey data that include activity reviews from their customers. The survey dataset that contains customer reviews will be the main ingredient for this text analytics project to create a comprehensive analysis that will answer these business questions: 1) What are the most frequent meaningful words used by customers that associated with positive and negative sentiments? 2) What deeper insights can be drawn from those frequently used words to enhance customer experience? 3) Which customers will most likely recommend the theme park to others?
This project will use several text analytics techniques and methods that have been discussed in class, such as tokenization and stop-word removal, to pre-process the data before the actual analytics process.
The analysis will contain single and bi-gram word-cloud visualizations to illustrate the most frequent meaningful words used by the customers in their reviews. The single word-cloud graphs will be grouped by their sentiments, whether positive or negative, for each of the different activities.
The most common words found on the word cloud visualizations will be analyzed carefully using word embeddings techniques. A word-embedding model will be trained to gain deeper insights from those frequently used words to provide recommendations for the management to enhance customer experience.
Lastly, the project will use a random forest algorithm to create a classification model from the text data to predict which customers will most likely recommend the theme park to others.
Comments