This document summarizes a project to reduce fraudulent card transactions for a US national bank. An ensemble technique using logistic regression and K-nearest neighbors was developed to classify transactions as fraudulent or legitimate in real time. The project was estimated to reduce fraudulent losses by $16-18 million while costing $4.2 million to develop. Testing on 1 year of transaction data accurately classified transactions and reduced fraudulent cases by 80-90%, saving the bank $16 million.
Credit Card Fraudulent Transaction Detection Research PaperGarvit Burad
Credit Card Fraudulent Transaction Detection Research Paper using Machine Learning technologies like Logistic Regression, Random Forrest, Feature Engineering and various techniques to deal with highly skewed dataset
Credit Card Fraudulent Transaction Detection Research PaperGarvit Burad
Credit Card Fraudulent Transaction Detection Research Paper using Machine Learning technologies like Logistic Regression, Random Forrest, Feature Engineering and various techniques to deal with highly skewed dataset
AlgoCharge offers a web-based fraud management system that assists in credit card fraud detection & prevention with Geo-based filters. The system provides various levels of fraud protection to enhance acceptance rate & reduce the risk of charge-backs.
Credit Card Fraud Detection Using ML In DatabricksDatabricks
In the Credit Card Companies, illegitimate credit card usage is a serious problem which results in a need to accurately detect fraudulent transactions vs non-fraudulent transactions. All organizations can be hugely impacted by fraud and fraudulent activities, especially those in financial services. The threat can originate from internal or external, but the effects can be devastating – including loss of consumer confidence, incarceration for those involved, even up to downfall of a corporation. Despite regular fraud prevention measures, these are constantly being put to the test in an attempt to beat the system.
Fraud detection is a task of predicting whether a card has been used by the cardholder. One of the methods to recognize fraud card usage is to leverage Machine Learning (ML) models. In order to more dynamically detect fraudulent transactions, one can train ML models on a set of dataset including credit card transaction information as well as card and demographic information of the owner of the account. This will be our goal of the project while leveraging Databricks.
Build an Ensemble classifier that can detect credit card fraudulent
transactions.Implemented a classifier by use of machine learning algorithms, such as
Decision Trees, Logistic Regression, Artificial Neural Networks and Gradient Boosting
Classifier.
Machine Learning (ML) for Fraud Detection.
- fraud is a big problem (big data, big cost)
- ML on bigger data produces better results
- Industry standard today (for detecting fraud)
- How to improve fraud detection!
Artificial Intelligence for Banking Fraud PreventionJérôme Kehrli
Artificial Intelligence at NetGuardians:
"From skepticism to large scale adoption towards fraud prevention"
Slides of my speech at the EPFL / EMBA Innovation Leader 2018 event.
This deck is from Interpol Conference 2017, these slides shows the holistic view of machine learning in cyber security for better organization readiness
This presenation shows how to deal with the problem of fraud detection with
1. Classic machine learning techniques. All supervised machine learning algorithms for classification will do, e.g. Random Forest, Logistic Regression, etc.
2. Techniques from the outlier detection or the anomaly detection approach, e.g. autoencoder and isolation forest
First presented by Kathrin Melcher (KNIME) at ODSC Europe in London in November 2019.
Dealing with the hassles of credit card fraud or identity theft can be frustrating and time consuming. This training will provide tips on how to protect yourself, your clients and your loved ones.
AlgoCharge offers a web-based fraud management system that assists in credit card fraud detection & prevention with Geo-based filters. The system provides various levels of fraud protection to enhance acceptance rate & reduce the risk of charge-backs.
Credit Card Fraud Detection Using ML In DatabricksDatabricks
In the Credit Card Companies, illegitimate credit card usage is a serious problem which results in a need to accurately detect fraudulent transactions vs non-fraudulent transactions. All organizations can be hugely impacted by fraud and fraudulent activities, especially those in financial services. The threat can originate from internal or external, but the effects can be devastating – including loss of consumer confidence, incarceration for those involved, even up to downfall of a corporation. Despite regular fraud prevention measures, these are constantly being put to the test in an attempt to beat the system.
Fraud detection is a task of predicting whether a card has been used by the cardholder. One of the methods to recognize fraud card usage is to leverage Machine Learning (ML) models. In order to more dynamically detect fraudulent transactions, one can train ML models on a set of dataset including credit card transaction information as well as card and demographic information of the owner of the account. This will be our goal of the project while leveraging Databricks.
Build an Ensemble classifier that can detect credit card fraudulent
transactions.Implemented a classifier by use of machine learning algorithms, such as
Decision Trees, Logistic Regression, Artificial Neural Networks and Gradient Boosting
Classifier.
Machine Learning (ML) for Fraud Detection.
- fraud is a big problem (big data, big cost)
- ML on bigger data produces better results
- Industry standard today (for detecting fraud)
- How to improve fraud detection!
Artificial Intelligence for Banking Fraud PreventionJérôme Kehrli
Artificial Intelligence at NetGuardians:
"From skepticism to large scale adoption towards fraud prevention"
Slides of my speech at the EPFL / EMBA Innovation Leader 2018 event.
This deck is from Interpol Conference 2017, these slides shows the holistic view of machine learning in cyber security for better organization readiness
This presenation shows how to deal with the problem of fraud detection with
1. Classic machine learning techniques. All supervised machine learning algorithms for classification will do, e.g. Random Forest, Logistic Regression, etc.
2. Techniques from the outlier detection or the anomaly detection approach, e.g. autoencoder and isolation forest
First presented by Kathrin Melcher (KNIME) at ODSC Europe in London in November 2019.
Dealing with the hassles of credit card fraud or identity theft can be frustrating and time consuming. This training will provide tips on how to protect yourself, your clients and your loved ones.
The Science of Predictive Maintenance: IBM's Predictive Analytics SolutionSenturus
Overview of IBM’s Predictive Maintenance and Quality (PMQ) solution. View the webinar video recording and download this deck: http://www.senturus.com/resources/science-predictive-maintenance/.
We show you the PMQ solution can keep manufacturing processes, infrastructure and field equipment running to maximize use and performance, while minimizing costs.
We show how you can use powerful analytics and data integration to help: Anticipate asset maintenance and product quality problems, Reduce unscheduled asset downtime, Spend less time solving production machinery and field asset problems, Improve asset productivity and process quality, Monitor how assets are performing in real-time and predict what will happen next.
Senturus, a business analytics consulting firm, has a resource library with hundreds of free recorded webinars, trainings, demos and unbiased product reviews. Take a look and share them with your colleagues and friends: http://www.senturus.com/resources/.
Leverage IoT to Setup Smart Manufacturing SolutionsSoftweb Solutions
The Internet of Things (IoT) is now to involve in manufacturing unit to deliver and enhance the productivity of companies through smart factory concept. It gives full business insights of manufacturing process and deliver data on their devices. View more at - http://www.softwebsolutions.com/iot-manufacturing-solutions.html
Chris Day presents at 1st ASCII MSP event of 2016 in San Marcos, TX. Presentation covers the value of great documentation and process and in-depth ROI review.
Managing an Experimentation Platform by LinkedIn Product LeaderProduct School
Main Takeaways:
-Establishing a culture of experimentation at scale
-Developing the product vision and strategy
-Backlog prioritization based on Impact Score formula
Business is always in a constant state of flux- more so these days, with disruption happening all around. How do you move from your AS IS state to TO BE architecture in your enterprise transformational journey? What mix and match of people, processes and technology will you blend together, and in what proportion, to drive enterprise value to deliver transformational results? TOGAF has a suite of tools that can help architects to chalk out the architectural roadmap for enterprise success. This talk will also focus on how agility is an underlying thread in this framework, and how value is delivered incrementally, making the process robust and
bankable.
Key Takeaways
Exposes the audience to the features of TOGAF which help plug business technology gaps.
How TOGAF has agility at its core to drive transformational results.
Why it is a good skill and knowledge for a seasoned IT professional to have in their kitty.
Digital Manufacturing and Design Innovation InstitutePlantEngineering
Announced earlier this year, the Digital Manufacturing and Design Innovation Institute (DMDII) is a Chicago-based manufacturing hub that will bring together public, educational and private interests to accelerate innovation and reduce development time and costs. Learn how all manufacturing will benefit from the research and development based at this digital lab.
Digital Manufacturing and Design Innovation InstituteControlEng
Announced earlier this year, the Digital Manufacturing and Design Innovation Institute (DMDII) is a Chicago-based manufacturing hub that will bring together public, educational and private interests to accelerate innovation and reduce development time and costs. Learn how all manufacturing will benefit from the research and development based at this digital lab.
In this part 6 of Fast Track Machine Learning (Machine Learning Overview) series, Dr. Dakshinamurthy Kolluru explains about the Big Data.
He explains about Big Data and how the issue is resolved using Big Data. He also explains what is Pig, Hive, Hadoop.
In this part 5 of Fast Track Machine Learning (Machine Learning Overview) series, Dr. Dakshinamurthy Kolluru explains about the different aspects which makes a data science problem tough.
He says that it is easy to work with structured data rather than an unstructured data and explains why it is so.
In part 4 of Fast Track Machine Learning (Machine Learning Overview) series, Dr. Dakshinamurthy Kolluru explains about the three main aspects that are to be given importance while defining the architecture in Machine Learning.
He explains about the difference between training, testing data and why is it important to keep testing data in a given data set.
In Part 3 of Fast Track Machine Learning (Machine Learning Overview) series Dr. Dakshinamurthy Kolluru explains that as a Machine Learning expert one has to give more importance to the 'Customer' rather than the way algorithm is developed.
Based on customer's requirement, finalize the output forms of knowledge. The form could be a rule or equation or graph or a black box.
In Part 2 of Fast Track Machine Learning (Machine Learning Overview) series Dr. Dakshinamurthy Kolluru explains that all Machine Learning can be treated as Pattern Search.
The 5 different searches in Machine Learning are:
1. Exhaustive Search
2. Random Search
3. Mathematical Search
4. Greedy Search
5. Guided Random Search
He explains all the five with the help of different real-world examples.
In part 4 of Fast Track Machine Learning (Machine Learning Overview) series, Dr. Dakshinamurthy Kolluru explains about the three main aspects that are to be given importance while defining the architecture in Machine Learning.
He explains about the difference between training, testing data and why is it important to keep testing data in a given data set.
In Part 3 of Fast Track Machine Learning (Machine Learning Overview) series Dr. Dakshinamurthy Kolluru explains that as a Machine Learning expert one has to give more importance to the 'Customer' rather than the way algorithm is developed.
Based on customer's requirement, finalize the output forms of knowledge. The form could be a rule or equation or graph or a black box.
In Part 2 of Fast Track Machine Learning (Machine Learning Overview) series Dr. Dakshinamurthy Kolluru explains that all Machine Learning can be treated as Pattern Search.
The 5 different searches in Machine Learning are:
1. Exhaustive Search
2. Random Search
3. Mathematical Search
4. Greedy Search
5. Guided Random Search
He explains all the five with the help of different real-world examples.
In part 1 of Fast track Machine learning (Machine Learning Overview) series Dr. Dakshinamurthy Kolluru gives a thousand feet view of Machine Learning.
One can divide the problems in Machine Learning as Classification, Regression and Optimization. Where in
Classification can be defined as splitting the space.
Regression is fitting a curve and regression can also be set up as a classification problem.
Optimization is on a curve, find the maximum and minimum points.
2. Domain & Challenges
Portfolio
Probability of Default
Allocation
Fraud Detection
Top Performing
Highest Value Customers
Agents
Churn
3. Business Problem
A US national bank which has a revenue of $10 billion, is
losing about 2% of it’s revenue, i.e $20 million, due to
fraudulent card transactions.
5. Consultation
Reduce the fraudulent cases by about 80-90%.
Losses curtailed: $16 - $18 million
Price of Information (Including Product Cost): $4.2
million
6. Data:
Approximately 1 year data
500,000 records
2% fraud and 98% legitimate
Attributes:
Location, Customer ID, Date, Time, Transaction
Amount, Account ID, Reference ID, Transaction
Code, Membership Period, Credit Card
Limit, Fraudulent Cases (Yes/No)
7. Architecture:
System 2
• Neural • K – Nearest
Networks • Logistic Neighbours
Regression
System 1 System 3
9. Cost Estimates:
3 machines, 1 shared memory
6 machines per state
1 server
Machine Cost, Server cost & Shared memory cost:
$100,000 – one time investment
Back up machines: 50 ~ $15,000
Server Maintenance Cost: $20,000 per year
Total Cost incurred: $115,000 one time + $20,000 per
year maintenance
10. Product
1 – 3 Scale rating
Aim to classify any new transaction as fraudulent or
not on the basis of the rating.
Any transaction with an average rating of 2.7 or more
is flagged “RED” indicating with more than 90%
evidence.
Alert sent to the Bank and Customer immediately.
Evaluation is done real time.
11. Product Pricing
2 months to analyse the data.
4 months to build models and test and improve.
Project Requires – 12 Analysts, 2 Managers
Cost To Company for employees: - $504,000 + $120,000
= $624,000
Additional Expenses approximately $300,000.
Price of building Product: Approx $924,000
12. Results:
$16 million saving!!
25
20
20
15
Initial Losses
10
After
4
5 implementation
0
Initial Losses After
implementation
13. International School of Engineering
2-56/2/19, Khanamet, Madhapur, Hyderabad - 500 081
For Individuals: +91-9177585755 or 040-65743991
For Corporates: +91-9618483483
Web: http://www.insofe.edu.in
Facebook: http://www.facebook.com/insofe
Twitter: https://twitter.com/INSOFEedu
YouTube: http://www.youtube.com/InsofeVideos
SlideShare: http://www.slideshare.net/INSOFE
LinkedIn: http://www.linkedin.com/company/international-
school-of-engineering
This presentation may contain references to findings of various reports available in the public domain. INSOFE makes no representation as to their accuracy or that the
organization subscribes to those findings.
The best place for students to learn Applied Engineering http://www.insofe.edu.in