Previous MSc Theses

2019 | 2017 | 2016 | 2015 | 2014
2013 | 2012

"Applying Business Analytics in Practice to a Bank Telemarketing Dataset." O. E. Oluwabusola. D. Roussinov. Department of Computer and Information Sciences, University of Strathclyde. 2016. Download PDF (BibTeX) IM

The business analytics approach to doing business involves the use of an organization's transactional data to gain knowledge on how business processes can be improved through the use of data mining techniques which are aimed at identifying interesting patterns that can be adopted by an organization to make more data-driven decisions.

As part of the process of driving the relevance of Business Analytics and Data Mining approach in businesses, this project focused on applying several data mining techniques to help identify patterns on how data gathered from a Portuguese bank telemarketing activity can be used to determine the likelihood of customers subscribing to term deposits. The dataset used contains 20 input attributes regarding information about the bank telemarketing campaigns conducted by a Portuguese bank and a target variable was used to predict if a customer would be subscribing to a term deposit.

When dealing with real world dataset such as that being used in this research, there is usually the problem of class imbalance where the occurrence of one class is more predominant than the other. To prevent the underperformance of the learning algorithms due to the imbalance class problem in this dataset, the research focused on adopting the data preprocessing method, the feature selection method and the use of ensembles to ensure that the classification algorithms chosen reach their optimum predictive performance.

The data mining task was carried out by using the Weka Machine learning tool to analyse the dataset. During the course of this research, it was discovered that the process of combining the feature selection task with the use of the ensembles learning method was the best approach to improving the predictive capabilities of the learning algorithms used. Association rules were also discovered in relation to analysing what factors contribute to the most likely reasons why customers would subscribe to the bank's term deposit.