| Popis: |
In identifying potential customers who would have a requirement for a loan by using direct marketing, data mining techniques come to our rescue. In order to identify potential customers from very large data, we need an algorithm that optimizes two parameters (i) high classification accuracy and (ii) minimum of error rates. In this paper, we propose a Kernel-Fold-based Confusion Matrix (KFCM) approach that when applied to existing Logistic Regression, Random Forest, SVM, AdaBoost, Stochastic Gradient and Naive Bayes, Data Mining Algorithms narrows down the list of potential customers who may have requirements for a loan. It has been observed that for Logistic Regression algorithm, there is a significant improvement in classification accuracy. In this paper, data set used is taken from the UCI Machine Learning Repository. |
