ETD Collection

Permanent URI for this collectionhttps://wiredspace.wits.ac.za/handle/10539/104


Please note: Digitised content is made available at the best possible quality range, taking into consideration file size and the condition of the original item. These restrictions may sometimes affect the quality of the final published item. For queries regarding content of ETD collection please contact IR specialists by email : IR specialists or Tel : 011 717 4652 / 1954

Follow the link below for important information about Electronic Theses and Dissertations (ETD)

Library Guide about ETD

Browse

Filters

20181

Settings

Author: search.filters.author.Maplanka, Ntombizodwa

Search Results

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    A study to use a data mining approach to classify customer price sensitivity using a retail banking foreign exchange historical dataset
    (2018) Maplanka, Ntombizodwa
    Data analysis combined with machine learning has become an essential part of the modern scienti c methodology, o ering automated procedures for the prediction of a phenomenon based on past observations, identifying underlying patterns in data and providing insights about the problem. This thesis seeks to demonstrate the use of data mining techniques to classify price sensitive customers using a historical dataset from a retail banking forex department. Two data sets were merged (customer data and deals data), and statistical models were tted and compared; namely decision trees, random forests and neural networks. All models produced excellent results when tted on the datasets; the random forests performed slightly better with marginal improvements over decision trees and neural networks. These models gave the area under the receiver operating characteristic curve of at least 0.90 and percentage correctly classi ed of least 0.95 for the datasets. Apart from making the most accurate predictions of the response variable random forests and decision trees were used to identify predictor variables that are most important to make these predictions. The study shows that in retail banking under the given setting, the foreign exchange division can price the clients appropriately and increase competitive edge by using data mining techniques to predict customers' price sensitivity to foreign exchange rates. The next step for the bank is to use these methods to retain the customers, increase revenue as well as make improvements in pricing where warranted.