Benefit Recommendation System: A Use Case for Pension Fund Data Analytics
Patipan Prasertsom, Intouch Kunakorntum, Atthavit Tulyathan, Nontawit Cheewaruangroj, Kanyawee Pornsawangdee, Napassawan Pasuthip · 2020
To promote the usage of a mobile application and its benefit offers, a benefit recommendation system was proposed, in which different models were combined to improve performance and robustness. First, high-potential users were identified with a classification model based on users' demographics. Then, a recommendation model proposed personalized benefits based on their demographics and overall usage trends of the userbase. This model aimed for new and returning users whose personal historical usage data was absent. The two models could be used together to attract new users as well as to encourage their benefit usage. Finally, for existing active users, a recommendation model that utilized users' behavioral and preference data from their usage records was designed to further improve users' engagement. Overall, our system was able to achieve high prediction performance, up to 89.15% when recommending benefit categories to users.