Author(s): Devi, Octaviani
Date: 2016
Persistent ID: http://hdl.handle.net/10362/19788
Origin: Repositório Institucional da UNL
Subject(s): Automobile insurance; Rule‐based clustering; K‐means; Clustering; Classification; Decision tree
Author(s): Devi, Octaviani
Date: 2016
Persistent ID: http://hdl.handle.net/10362/19788
Origin: Repositório Institucional da UNL
Subject(s): Automobile insurance; Rule‐based clustering; K‐means; Clustering; Classification; Decision tree
Internship Report presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics
Defining pricing strategy is a challenge for every insurance company. Competition makes insurers need to be more careful to adjust the premium since it may affect the reaction of the existing customer or the new ones. Correspondingly, it may impact the relationship with customer, also the profitability of the company. Moreover, the increment of number of policies will lead to the diversity of policy’s risk profile and characteristics which becomes a challenge for insurer to manage their portfolio. Therefore, a deep understanding of portfolio segmentation is important for the company to fine tune the pricing strategy and gain more profit. The project aims to discover portfolio clusters by using k‐means clustering algorithm and extract the rules of each cluster by developing classification model using Decision Tree algorithm. The result of the model shows that the clusters give different characteristics and behavior. Complement with KPI metrics, the company is able to monitor the performance of each clusters. So that, the company may use the analyses to optimize the strategy of growth and profitability.