0% found this document useful (0 votes)
392 views2 pages

Poster

Uploaded by

Jos Buttler
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
392 views2 pages

Poster

Uploaded by

Jos Buttler
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

Introduction

In credit card purchases, "fraud" refers to the unauthorized and unwanted use of an account by
someone who is not the account's owner. Necessary preventive measures should be implemented in
order to stop this exploitation and to reduce and discourage such fraudulent activities in future. Fraud
detection entails tracking user behaviors in order to estimate, interpret, or prevent objectionable
behavior such as fraud, interference, and defaulting.

So solving this problem require the attention of ML and data science communities which can help build
an automated solution to this problem. This problem is challenging for learning perspective, because of
various factors such as class imbalance which means the fraudulent transactions are far outnumbered
by valid ones and statistical patterns of transactions also change over time.

Another problem is that in real world massive transaction requests are made at one time and to quickly
scan all of them an automated system is needed which can analyze all the authorized transactions and
report the suspicious ones.

Machine learning algorithms can use the historic data to learn from them by finding different patterns in
data and then using those patterns to classify the new transactions as valid or fraudulent. We’ll be using
ML algorithms to create models which can help detect the fraudulent transaction and also different
methods will be used to tackle the imbalanced data.

Methodology

In this project we'll use the latest machine learning algorithms will be built to tackle the challenge of
detecting anomalous activities and compare the different models performance.

First and foremost, we obtained our dataset from Kaggle.

There are 31 columns in this dataset, 28 of which are labelled v1-v28 to protect confidential data.

We used graphs to look for anomalies in the dataset and to visually understand it.

We implemented different Machine algorithms (Random Forest, Decision Tree) and a Deep Neural
Network. First we implemented it on the imbalance dataset and then by balancing the data using
different techniques.

Results:

The result of the different ML algorithms was calculated using different metrics of Confusion Matrix such
as Recall, F1 Score and Precision. And DNN performed better than all of the other on the Oversampled
Data.

Discussion
Although we did not achieve our target of 100 percent accuracy in fraud detection, we did create a
system that, with enough time and data, can come very near. As with any project of this nature, there is
space for improvement here.

The dataset contains a lot of room for improvements. As previously shown, the precision of the
algorithms increases as the size of the dataset increases. As a result, more data will undoubtedly
improve the model's accuracy in detecting fraud and reduce the number of false positives. This,
however, can only be achieve by backing from the banks themselves.

References
Andrea Dal Pozzolo, Olivier Caelen, Reid A. Johnson and Gianluca Bontempi. Calibrating Probability
with Undersampling for Unbalanced Classification. In Symposium on Computational Intelligence and
Data Mining (CIDM), IEEE, 2015

Carcillo, Fabrizio; Le Borgne, Yann-Aël; Caelen, Olivier; Bontempi, Gianluca. Streaming active


learning strategies for real-life credit card fraud detection: assessment and
visualization, International Journal of Data Science and Analytics, 5,4,285-300,2018,Springer
International Publishing

Bertrand Lebichot, Yann-Aël Le Borgne, Liyun He, Frederic Oblé, Gianluca Bontempi Deep-Learning
Domain Adaptation Techniques for Credit Cards Fraud Detection, INNSBDDL 2019: Recent
Advances in Big Data and Deep Learning, pp 78-88, 2019

Fabrizio Carcillo, Yann-Aël Le Borgne, Olivier Caelen, Frederic Oblé, Gianluca Bontempi Combining
Unsupervised and Supervised Learning in Credit Card Fraud Detection  Information Sciences, 2019

Acknowledgements

You might also like