The Role of Detection Rate in MAPE to Improve Measurement Accuracy for Predicting FinTech Data in Various Regressions

Abstract

Prediction is included in the data mining process to predict future data based on learning from past data. Various techniques are used in making predictions. The Regression method also includes techniques for making predictions. Various regressions such as Linear Regression, Ridge Regression, Lasso Regression, and Multivariate Adaptive Regression Splines (MARS) are regression techniques that are fond of being used in predicting data in business. Every prediction is always measured success with several formulations. As MAPE is a measuring tool in obtaining accuracy, so it is trying to be designed with the role of Detection Rate (DR) in order to get a smaller error value in obtaining accuracy. In this paper, the process of obtaining accuracy in Linear Regression is carried out to obtain a MAPE of 0.15874361801345002 % and the role of DR in MAPE is 0.1410249900632677 %. At Ridge Regression get a MAPE of 0.15820461185453846 % and the role of DR in MAPE is 0.14077739389387 %. On Lasso Regression get a MAPE of 0.14793925681569248 % and the role of DR in MAPE is 0.1370143839961479 %. On MARS get a MAPE of 0.16209808399129746 % and the role of DR in MAPE is 0.14528079908718253 %.