Analisis Perbandingan Metode Pohon Keputusan Quest dan Chaid Dalam Menilai Kinerja Bank Umum Konvensional

Abstract

The performance of conventional commercial banks can be determined from several parameters included in the RGEC analysis (Risk Profile, Good Corporate Governance, Earnings, Capital). This is done to help all parties such as customers, shareholders, and regulators understand the extent of the bank's stability, health and capability in dealing with various economic situations. In addition, this analysis also supports investment decision making, banking sector supervision and maintaining public confidence in the banking industry. Indicators in the RGEC analysis used include Capital Adequacy Ratio (CAR), Return on Assets (ROA), Net Interest Margin (NIM), Non-Performing Loan (NPL), Loan to Deposit Ratio (LDR), Operating Costs to Operating Income (BOPO), NPL Level, ROA Level and Bank Rating according to the Indonesian Securities Rating Agency (PEFINDO). The decision tree method used in this comparative analysis is QUEST (Quick, Unbiased, Efficient, and Statistical Tree) which will produce a binary classification tree and CHAID (Chi-squared Automatic Interaction Detection) which will produce a non binary classification tree. The results showed that QUEST produces indicators of BOPO, ROA, NPL and Bank Rating while CHAID produces indicators of NPL, ROA, LDR, and BOPO as the most significant indicators in predicting the performance of conventional commercial banks. The results of the indicator analysis produced by the two decision tree methods will be compared and analyzed in detail using the confusion matrix.