THE MODELLING OF INTERVALS BETWEEN INCIDENT AND REPORTING TIME ON FIRE INSURANCE DATA USING MULTIPLE LINEAR REGRESSION AND COKRIGING

• Main Author: Prapti Ningrum, Ira
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/47805
• Institution: Institut Teknologi Bandung

Intervals between incident and reporting time is the time difference between the time of reporting a claim to the insurance company and the time of the fire incident. In insurance research rarely uses spatial analysis methods especially about the intervals between incident and reporting time. One method of spatial analysis is cokriging. Cokriging is used to determine the estimated value of the intervals between incident and reporting time.Before using the cokriging method, an analysis is carried out to determine the greatest effect of the total coverage, the total claims and the distance to the intervals between incident and reporting time, multiple linear regression analysis was used. In multiple linear regression, a model with a matching model of 83.3% was produced and the variable that most influences the intervals between incident and reporting time is the total claim. By doing a logarithmic transformation of the total claims, the cokriging method is performed by analyzing the isotropic semivariogram and cross-semivariogram models for intervals between incident and reporting time and total claims. Semivariogram models used in this study are exponential, gaussian, and spherical models. In the spherical model cokriging gives the smallest number of error squares which is 10,316.41 and RMSP (Root Mean Square Standardized Prediction Error) of 0.838 which means cokriging with the spherical semivariogram model gives a fairly reliable estimation validity. Cokriging with a spherical semivariogram model is used to determine the estimated intervals between Incident and reporting time and total claims.