چکیده:
The automotive body insurance field, contrary to what it seems, is not considered a very profitable field for insurance companies and is moving towards loss-making. Therefore, the present research pays attention to the adequacy of automotive body insurance rates and measures for them, and seeks to improve and scientize these matters as much as possible. Accordingly, first, by identifying the variables influencing the risk of policyholders and matching the variables with the data available in the database of the studied company, the final factors were selected; Then, while performing preprocessing operations on the data using the neural network model, the damage category and the amount of potential damage of policyholders were predicted so that insurance companies can define optimal rates for policies by considering them and their desired damage coefficient. The results of the research show that the presented model can estimate the damage category with 91% accuracy and predict the amount of potential damage of policyholders with 87% accuracy.
خلاصه ماشینی:
Data mining stages based on the CRISP-DM standard Empirical Background of the Research Yeo, Smith, Willis, and Brooks (2001) used clustering techniques and considered demographic criteria such as driver age and gender to cluster customers and their related risk; then, in each cluster, using a neural network model and considering the aforementioned criteria, they predicted changes in the insurance premium.
Anbari, Nadali, and Islami Nasr Abadi (1389) in a research considering the characteristics of the insurer such as his age and gender, along with the characteristics of the car such as its use and age, compared several models (decision tree, neural networks, Bayesian networks, support vector machine, logistic regression, discriminant analysis) to predict the damage class of insurers and categorized them into three classes: low risk, medium risk, and high risk.
Final factors selected after consulting with experts Insurer Characteristics: Age, gender, marital status, city of residence, year of obtaining a license (driving experience), level of education, income level, occupation, number of claims in the previous year, distance between work and home Car Characteristics: Type, color, year of manufacture, safety equipment (ABS), mileage in kilometers, usage, group, license plate type, number of cylinders, type, current value Insurance Policy Characteristics: Amount of insurance coverage Data Collection This research used data available in the database of automobile body insurance of one of the domestic insurance companies (Asia Insurance).
(
, 
, 
)
