- Balkan Journal of Electrical and Computer Engineering
- Vol: 3 Special Issue
- Establishing the Potential Clients Using Artificial Neural Networks
Establishing the Potential Clients Using Artificial Neural Networks
Authors : Ziynet Pamuk, Yüksel Yurtay, Oğuz Yavuzyilmaz
Pages : 219-224
View : 24 | Download : 7
Publication Date : 2015-12-30
Article Type : Research
Abstract :Today, technologies retrieving forward-looking information from the existing data are available. In this study, whether the clients would open a deposit account was estimated using the data in the marketing campaign of a bank in Portugal for its clients. The purpose of the study was to create a decision support system to determine the potential clients in future. The data set collected from 4.512 subjects consists of 16 input attributes (job, age, balance, etc.) and 1 output attribute (yes/no). In the study, the 6-fold cross validation method was used. The data obtained from 3.760 people were used for the training process and the data obtained from 752 people were used for the testing process. As classifiers; Feed Forward Neural Networks (FFNN), Probabilistic Neural Network (PNN) and k Nearest Neighbor (kNN) were used. At the end of the study, success ratios of different algorithms were compared by Receiver Operating Characteristics (ROC) analysis method. Feed forward neural network yielded the best result with an accuracy rate of 95.74%.Keywords : Bank data, deposit, artificial neural network, k nearest neighbor algorith