Boosting Indonesian Telecom's Revenue with Recommender System for Specific Customer Segmentation

This research explores a new approach to building recommendation systems in the telecommunications industry, focusing on revenue generation rather than traditional service ratings. By analyzing data from a well-known local telecommunications company, this research proposes two different strategies: one targeting customers who generate minimal revenue and another for customers who have higher and consistent revenue. Utilizing collaborative filtering with item-based techniques, the system uses Cosine and Jaccard Similarity algorithms to improve its recommendation results. The Jaccard Similarity algorithm, in particular, showed a high accuracy rate of 93%. This method, which is based solely on income data and customer segmentation based on average daily income, offers new insights into the development of recommendation systems in the absence of explicit service ratings.