APPLICATION OF DATA MINING TECHNIQUES FO R CUSTOMER SEGMENTATIONS AND PREDICTION: CASE SINKE BANK, ASSELA BRANCH

Abstract

Identifying customers who are more likely to respond positively to a product or service offering is an important issue in business decision-making. In customer identification, data mining has been widely used to predict potential customers for various products and services. The main goal of this thesis is to develop a model that classifies customers for Sinke Bank. Since there were no predefined classes describing the bank’s customers, the researcher applied clustering techniques to determine an appropriate number of customer segments. Subsequently, a predictive model was developed to identify potential customers, achieving an accuracy of 99%. For modeling purposes, data was collected from the institution’s head office. Because irrelevant features can negatively affect model performance, data preprocessing was conducted to identify the most relevant inputs. Various data mining techniques and algorithms were applied throughout the modeling process to address related challenges effectively. The K-means clustering algorithm was used to segment customer records into groups with similar characteristics. Different parameters were tested before identifying a segmentation that made sound business sense. The J48 decision tree algorithm was then used for classification. In addition to attributes identified by domain experts as highly influential for customer segmentation, the loan amount attribute was found to have a significant impact. Overall, the findings of this study are encouraging and demonstrate the potential application of data mining solutions in the banking industry, particularly in customer segmentation and prediction for Sinke Bank Corporation.