Combining Predictions of Auto Insurance Claims

Abstract

This paper aims at achieving better performance of prediction by combining candidate predictions, with the focus on the highly-skewed auto insurance claim cost data. We analyze a version of the Kangaroo Auto Insurance company data, and incorporate different combining methods under five measurements of prediction accuracy. The results show: 1) When there exists an outstanding prediction among the candidate predictions, the phenomenon of the "forecast combination puzzle" may not exist. The simple average method may perform much worse than the more sophisticated combining methods; 2) The choice of the prediction accuracy measure is crucial in defining the best candidate prediction for "low frequency and high severity" datasets. For example, Mean Squared Error (MSE) does not distinguish well different combining methods, the MSE values of which are very close to each other; 3) The performances of different combining methods can be drastically different. 4) Overall, the combining approach is valuable to improve the prediction accuracy for insurance claim cost.

Tweedie Gradient Boosting for Extremely Unbalanced Zero-inflated Data

“…Gini index is a well-accepted tool to evaluate the performance of the predictions, especially for extremely zero-inflated data set. There exists many variants of Gini index and one variant we use is proposed by Ye et al (2018), denoted by Gini a . For a se-…”

“…In this simulation study, we investigate and compare the performance of TDboost, ZIF-TDboost and Grabit in terms of MAD(w.r.t. true losses and expected losses), RMSD , Re RMSD and two variants of Gini index proposed by Ye et al (2018) and Frees et al (2011). We consider the data generated from the Zero-Inflated Tobit (Tobin, 1958) model where the true target function is given by…”