Perbandingan Algoritma Klasifikasi K-Nearest Neighbor, Random Forest dan Gradient Boosting untuk Memprediksi Ketertarikan Nasabah pada Polis Asuransi Kendaraan

An insurance policy provides coverage for compensation for specified loss, damage, illness, or death in exchange for premium payments. Likewise for vehicle insurance, every year the customer needs to pay a premium to the insurance company so that if an accident occurs that is not profitable for the vehicle, the insurance company provides compensation to the customer. The purpose of this research is to classify the health insurance cross-sell prediction dataset so that certain patterns or relationships can be found between the data to become valuable information and build a model to predict whether policyholders (customers) from the previous year will also be interested in insurance. Vehicles provided by the company. The researcher uses the K-nearest neighbor classification algorithm, Random Forest, and gradient boosting classifier as well as Python data mining tools. After doing the research, it was found that the K-nearest neighbor classification algorithm produces a higher accuracy of 91%, when compared to the Random Forest algorithm which is 87% and the boosting classifier algorithm is 88% in classifying customer interest in taking a vehicle insurance policy