Classification Method of Financial Behaviour Through Means of Machine Learning : Can a classification method created using bank transaction and machine learning help individuals to understand their spending behavior?

Sammanfattning

With the current fast transformation from physical cash to digitized banking systems, there are more and more people that are at risk of overspending without realizing it. There are methods and researches done that are targeted at incorporating machine learning in identifying fraudulent transactions and credit scores but currently there is no research done in categorizing people’s behaviour based on transaction records using machine learning techniques. The purpose of this project is to create such classification method based on income and spending. We propose a classification model based on k-means clustering and neural network techniques that categorizes people’s spendings behaviour based on monthly transaction records. The goal of this work is to contribute to one’s understanding of personal spending behaviour. The research question posed to achieve this is: can a classification method created using bank transaction and machine learning help individuals to understand their spending behaviour? This work is intended to be a basis for future research in studies in spending behaviour classification. The result shows the final method consists of a pipeline of simplifying bank transaction dataset, k-means, data augmentation and artificial neural network. It is capable of classifying spending behaviours based on characteristics such as ”low income and high spending” or ”high income and low spending”. The general approach for research method of this thesis is qualitative and inductive