Predictive Modeling of Claims in Flood and Mobile Home Insurance using Machine Learning
DOI:
https://doi.org/10.70179/vj8r7755Keywords:
Predictive modeling, flood insurance, mobile home insurance, machine learning, claim prediction, risk assessment, loss forecasting, insurance analytics, supervised learning, regression models, classification models, claims data, policyholder data, geographic risk, weather data, catastrophe modeling, model accuracy, data preprocessing, feature engineering, model training, model validation, random forest, gradient boosting, neural networks, XGBoost, claim severity, claim frequency, damage prediction, insurance technology, actuarial modeling, risk mitigation, spatial analysis.Abstract
Specific modeling of insurance claims is of utmost importance to any insurer for improving the decision making in pricing, reserving, and risk selection. In this work, we have undertaken predictive modeling of claims by developing algorithms using different classification techniques. The predictive modeling is carried out for two distinct classes of property insurance – flood insurance and mobile home insurance. Flood insurance business is characterized by highly concentrated losses, thus there is a need to better understand the underlying pattern of claims magnitude which could help insurers to transfer the risk. In our study, we show that proximity to flood zone and loss history in the counties where the agent practices has a significant impact on predicting claims. Personal mobile manufacturing of mobile homes creates a high concentration of fire claims; hence, well-timed and periodic reviews are needed to analyze the estimates on fire mobilization and direct residualer on claims. In our study, we show that applicants with higher insurance amount and a manufactured mobile home are supposed to submit more claims along with an increasing insurance value and construction year which is directly applicable to estimated loss amount and claims loss ratio. Also, the modeling study will guide insurers to determine the time and money needed for a potential applicant to complete the risk review process.
