Institutional Repository [SANDBOX]
Technical University of Crete
EN  |  EL

Search

Browse

My Space

Machine learning algorithms for customer acquisition: a comparative evaluation in banking services

Tavernaraki Maria-Zouzanna

Full record


URI: http://purl.tuc.gr/dl/dias/AB63FE3E-E1BC-47F3-8D64-E5F4698C8D11
Year 2020
Type of Item Diploma Work
License
Details
Bibliographic Citation Maria-Zouzanna Tavernaraki, "Machine learning algorithms for customer acquisition: a comparative evaluation in banking services", Diploma Work, School of Production Engineering and Management, Technical University of Crete, Chania, Greece, 2020 https://doi.org/10.26233/heallink.tuc.86873
Appears in Collections

Summary

The knowledge of the customers that a bank and every business possess is an important step in order to be able to predict the customers behavior that it wants to attract. Knowing the preferences of customers, it is possible to predict the preferences of others with similar behavior. One of the most common methods used by companies for this purpose is the use of machine learning algorithms. The purpose is to guide the reader through the analysis of the value of customer relationships and the crucial contribution of data mining tools and techniques to the application of machine learning algorithms. The algorithms try to find the banking products of a Spanish bank that are more likely to be bought and by which customers based on the consumers behavior and the behavior of others that already bought these products. To solve the problem, classification models were used which were executed through the R program. The methods used are: Logistic Regression, Extreme Gradient Boosting, Classification and Regression Trees, Boosted Logistic Regression and Neural Networks. Finally, the models created were evaluated through the AUROC index and their classification was performed based on their performance. For each model, the significance of the variables used to extract the results was calculated.

Available Files

Services

Statistics