Comparative result of credit approval with 3 models of machine learning algorithm using program R

  • สุเมธ จุฑาจันทร์ มหาวิทยาลัยหอการค้าไทย
  • สมพร ปั่นโภชา
Keywords: Credit Approval Analysis, Machine Learning, KNN (k-Nearest Neighbors), Accuracy Score, Precision Score, Recall, F1-Score


At present of financial innovations,  data scientists   have  used  Artificial Intelligence(AI )  to  analyze big data with  Machine Learning ( ML)  for serve customers to  more convenient and faster  such as using AI to analyze credit approval  that acts on behalf of  the financial officer  or  to analyze  credit card  emergency loan approval etc.          Due to limited access to and use of the data in Thailand. So  the objective of  this study was to  study  machine learning  in  financial data and  measure  efficiency of   3 ML algorithms of  the Credit Approval Analysis  by using  German Credit data, from the UC Irvine Machine Learning Reposity, and using R Programming as a tool to create Machine Learning.

The results showed that the KNN  model  had  the most efficient, with Accuracy Score, Precision Score, Recall and F1-Score as  99.33%, 99.03%, 100% and 99.51%, respectively, followed by  Logistic Regression simulated with Accuracy Score, Precision Score, Recall and F1-Score as 73.00%, 75.00%, 90.73% and 82.12% respectively, whereas CART Model with Accuracy Score 69.00%, Precision Score 87.32%, Recall 72.76% and F1-Score 79.38%.


Most read articles by the same author(s)

1 2 3 4 5 > >>