Data driven recommendation of personal insurance policies

• Main Author: Heng, Sabrina Chor Chen
• Other Authors: Sourav S Bhowmick
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/166034
• Institution: Nanyang Technological University

With the advancement of technology, Singapore remains as one of the most digitally competitive country in the world. With such progress as a technology hub, this had also led to the increase in the median gross monthly salary of fresh graduates. The shifting norms of neglecting insurances by most in the past has also changed to the younger ones seeing the importance of them. However, the average sales process of purchasing a policy that includes prospecting and advisory takes about one to two hours with prospecting taking about thirty minutes. The key aim of this project is to shorten the whole advisory process, mainly prospecting, from thirty minutes to a five-minute session while still ensuring that the needs and wants of the person are met. In doing so, this report will take one through the design, development, and data-analysis of an insurance Webform where users can answer a short series of questions, and a recommendation of insurance policies will be generated. The recommendation will be based on two things – inputs of users, and the usage of past users’ insurance data – to generate the policies recommended. To do so, a comprehensive study must first be done on the current solutions of the finance industry. We will then look at how this recommendation system helps bridge the current gap and provide a better efficient process. We will then look into the criteria of what makes an ideal recommendation system based on different individuals and what are some of the important questions to include in the webform before generating a personalised recommendation. The recommendation will be split into two parts, the first we will be going through how user inputs affect recommendations, and the second is based on past trends data, which will affect the recommendations to the first part.