Customer Profile Prediction model based on classification through approach Support Vector Machine (SVM)

Abstract

Nowadays the market is characterized globally, products and services are almost identical and there are many suppliers. The most important aspect in classifying data in data mining is classification. Classification techniques have been widely used in many problems in research. The purpose of this research is to build a model that can predict behavior based on the information of each customer. This research was conducted by making a Prediction Model of Customer Profile Based on Classification Through the Support Vector Machine Approach which aims to obtain a package prediction accuracy value that is suitable for WO (Wedding Organizer) customers in classifying based on the profile of prospective customers. In the optimization results on the SVM model kernel function, the linear and polynomial kernels get the same accuracy value on the training data of 99.29% and the testing data of 94.92%. The lowest accuracy value was obtained in the RBF kernel function of 97.16% on training data and 96.61% on testing data. the best precision class value in the data testing was obtained in the basic package at 100%. The total value of the appropriate prediction on the training data was obtained by 56 samples from a total of 59 samples, and 3 samples that did not match the prediction with an accuracy of 94.92% on the data testing

GS Cited Analysis