Text
Financing-Limit Prediction Classifier in Islamic Bank Using Tree-Based Algorithms
Objectives : This study aims to propose a machine learning algorithm
that could predict financing-limit based on customer
classification in Islamic bank.
Methods : The study took place in BJB Syariah bank during June -
July 2023. The analysis technique in this study is using
decision tree and random forest algorithms to find the
variables that best predict customer’s classification.
Results/findings : The results shows that the basic algorithm which is
Decision Tree gives 86% prediction accuracy. The algorithm
is then improved by using Random Forest algorithm. The
Random Forest algorithm gives 91% prediction accuracy
which significantly improves the base learning algorithm.
This study also shows that the higher accuracy, the better the
algorithm at capturing significant variables in the data.
Based on the experiments conducted, the variables that best
predict financing-limit category for a customer are
equivalent rate contract, type of business owners, amount of
financing needed, business revenue, amount of front
payment, financing term, customer type, collateral value,
economic sector, business code, type of receivables,
portfolio category, type of contract, category segmentation,
type of business, profit sharing ratio
No other version available