• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.17 no.4
• /
• pp.221-228
• /
• 2024

Predicting insurance claims is a key task for insurance companies to manage risks and maintain financial stability. Accurate insurance claim predictions enable insurers to set appropriate premiums, reduce unexpected losses, and improve the quality of customer service. This study aims to enhance the performance of insurance claim prediction models by applying ensemble learning techniques. The predictive performance of models such as Random Forest, Gradient Boosting Machine (GBM), XGBoost, Stacking, and the proposed Dynamic Weighted Ensemble (DWE) model were compared and analyzed. Model performance was evaluated using Mean Absolute Error (MAE), Mean Squared Error (MSE), and the Coefficient of Determination (R²). Experimental results showed that the DWE model outperformed others in terms of evaluation metrics, achieving optimal predictive performance by combining the prediction results of Random Forest, XGBoost, LR, and LightGBM. This study demonstrates that ensemble learning techniques are effective in improving the accuracy of insurance claim predictions and suggests the potential utilization of AI-based predictive models in the insurance industry.

• Asia pacific journal of information systems
• /
• v.34 no.2
• /
• pp.565-584
• /
• 2024

More than 1 million automobile insurance repairs occur per year globally, and the related repair costs add up to astronomical amounts. Insurance companies and repair shops are spending a great deal of money on manpower every year to claim reasonable insurance repair costs. For this reason, promptly predicting insurance claims for vehicles in accidents can help reduce social costs related to auto insurance. Several recent studies have been conducted in auto insurance repair prediction using variables such as photos of vehicle damage. We propose a new model that reflects auto insurance repair characteristics to predict auto insurance repair claims through an association rule method that combines gradient descent and location information. This method searches for the appropriate number of rules by applying the gradient descent method to results generated by association rules and eventually extracting main rules with a distance filter that reflects automobile part location information to find items suitable for insurance repair claims. According to our results, predictive performance could be improved by applying the rule set extracted by the proposed method. Therefore, a model combining the gradient descent method and a location-based association rule method is suitable for predicting auto insurance repair claims.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2015.11a
• /
• pp.60-61
• /
• 2015

In recent years, the accidents in construction projects are continuously increasing due to their complexity and variety. However, few studies have been conducted regarding the risk prediction model and the database of risk assessment in construction projects. To address of these issues, the objective of this study is to analyze the accident causes by using insured claim payouts of insurance companies. First, the descriptive analysis of accidents causes is conducted according to scheduling rate, season, and total construction costs. Second, the correlation analysis is conducted between accidents causes and total construction costs. In the future, the risk assessment model can be developed to quantify the accident causes in construction projects to estimate claim payouts of insurance companies.

• The Korean Journal of Applied Statistics
• /
• v.34 no.5
• /
• pp.697-710
• /
• 2021

Relative error prediction is preferred over ordinary prediction methods when relative/percentile errors are regarded as important, especially in econometrics, software engineering and government official statistics. The relative error prediction techniques have been developed in linear/nonlinear regression, nonparametric regression using kernel regression smoother, and stationary time series models. However, random effect models have not been used in relative error prediction. The purpose of this article is to extend relative error prediction to some of generalized linear mixed model (GLMM) with panel data, which is the random effect models based on gamma, lognormal, or inverse gaussian distribution. For better understanding, the real auto insurance data is used to predict the claim size, and the best predictor and the best relative error predictor are comparatively illustrated.

• International conference on construction engineering and project management
• /
• 2015.10a
• /
• pp.192-194
• /
• 2015

The growing size and complex process in construction project recently leads to increase risk and the losses as well. Even though researchers have identified the major risk indicators, there is lack of comprehensive and quantitative research for identifying the relationship between the risk indicators and economic losses associated with construction projects. To address this shortage of research, this study defines risk indicators and create a framework to assess the influence of economic losses from the indicators. An insurance company's claim payout record was accepted as the dependent variable to reflect the real economic losses. Based on the claims, we categorized the causes and results of accidents. To establish framework, built environment vulnerability indicators and geographical vulnerability indicators were employed as the risk indicators. A Pearson correlation analysis was adopted to validate the relationship with loss ratio and risk indicators. Consequently, this framework and its results may offer significant references for under writers of insurance companies and loss prevention activities.

• Journal of Korea Society of Industrial Information Systems
• /
• v.28 no.2
• /
• pp.1-9
• /
• 2023

Research using AI and big data is also being actively conducted in the health and medical fields such as disease diagnosis and treatment. Most of the existing research data used cohort data from research institutes or some patient data. In this paper, the difference in the prediction rate of survival and the factors affecting survival between breast cancer patients in their 40~50s and other age groups was revealed using health insurance review claim data held by the HIRA. As a result, the accuracy of predicting patients' survival was 0.93 on average in their 40~50s, higher than 0.86 in their 60~80s. In terms of that factor, the number of treatments was high for those in their 40~50s, and age was high for those in their 60~80s. Performance comparison with previous studies, the average precision was 0.90, which was higher than 0.81 of the existing paper. As a result of performance comparison by applied algorithm, the overall average precision of Decision Tree, Random Forest, and Gradient Boosting was 0.90, and the recall was 1.0, and the precision of multi-layer perceptrons was 0.89, and the recall was 1.0. I hope that more research will be conducted using machine learning automation(Auto ML) tools for non-professionals to enhance the use of the value for health insurance review claim data held by the HIRA.

• Journal of the Korea Institute of Building Construction
• /
• v.18 no.2
• /
• pp.151-159
• /
• 2018

Accident rate in the construction industry of South Korea is increasing every year, and it represents the highest percentage among industries. This shows that activities performed to prevent safety accidents in the country are not efficient when it comes to reduce the accident rate. In order to resolve this issue, a model for the prediction of human accidents should be established. In addition, it is required a quantification study based on pattern of human accidents. Therefore, the objective of this study is to quantify uncertainty of human accidents risk and predict how to change in various circumstances by using Monte Carlo Simulation. To achieve the objective, first, pattern of human accidents was defined. Second, insurance claim payout and information of human accidents during 14 years in construction site were collected. Third, descriptive analysis is conducted to determine the characteristics of the accident pattern. Fourth, to quantitatively analyze the pattern of the human accidents, the population of each accident occurrence and payout were estimated. Finally, estimated populations was analyzed according to characteristics of distribution by using Monte carlo simulation. In the future, this study can be used as a reference for developing the safety management checklist in construction site and development of prediction models of human accident.

• Journal of Preventive Medicine and Public Health
• /
• v.42 no.2
• /
• pp.135-142
• /
• 2009

Objectives : An appropriate sampling strategy for estimating an epidemiologic volume of diabetes has been evaluated through a simulation. Methods : We analyzed about 250 million medical insurance claims data submitted to the Health Insurance Review & Assessment Service with diabetes as principal or subsequent diagnoses, more than or equal to once per year, in 2003. The database was re-constructed to a 'patient-hospital profile' that had 3,676,164 cases, and then to a 'patient profile' that consisted of 2,412,082 observations. The patient profile data was then used to test the validity of a proposed sampling frame and methods of sampling to develop diabetic-related epidemiologic indices. Results : Simulation study showed that a use of a stratified two-stage cluster sampling design with a total sample size of 4,000 will provide an estimate of 57.04%(95% prediction range, 49.83 - 64.24%) for a treatment prescription rate of diabetes. The proposed sampling design consists, at first, stratifying the area of the nation into "metropolitan/city/county" and the types of hospital into "tertiary/secondary/primary/clinic" with a proportion of 5:10:10:75. Hospitals were then randomly selected within the strata as a primary sampling unit, followed by a random selection of patients within the hospitals as a secondly sampling unit. The difference between the estimate and the parameter value was projected to be less than 0.3%. Conclusions : The sampling scheme proposed will be applied to a subsequent nationwide field survey not only for estimating the epidemiologic volume of diabetes but also for assessing the present status of nationwide diabetes control.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2017.11a
• /
• pp.195-196
• /
• 2017

The number of human accidents in the construction industry is increasing every year, and it constitute the highest percentage among industry. This means that activities performed to prevent safety accidents in the country are not efficient to reduce the rate of accidents in the construction industry. In order to solve this issue, research has been conducted from various perspectives. But, research regarding to quantification model of human accidents is insufficient. the objective of this study is to conduct a basic study on quantification model development of human accidents. To achieve the objective, first, Cause of accident is defined the through literature review. Second, a basic statistic analysis is conducted to determine the characteristics of the accident causes. Third, the analysis is conducted after dividing into four categories : accumulate rate, season, total construction cost, and location. In the future, this study can be used as a reference for developing the safety management checklist for safety management in construction site and development of prediction models of human accident.