![National
Institute
of
Science
&
Technology
[1]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Credit Card Fraud Detection
Suraj Patro
CSE - 201710455
1701202348
M. Binayak Kumar Reddy
CSE - 201710558
1701202192
By](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-1-320.jpg)
![National
Institute
of
Science
&
Technology
[2]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Contents :
❑ Introduction
❑ Objective
❑ Implementation
❑ Software Tools
❑ Possible Outcome
❑ Performance Analysis
❑ Reference
❑ Conclusion](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-2-320.jpg)
![National
Institute
of
Science
&
Technology
[3]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Introduction :
Fraud is crime committed against property, involving the
unlawful conversion of the ownership of property to one's
own personal use and benefit.
Some Challenges in Fraud Detection :
❑ uncommon / very few in number
❑ concealed activities
❑ changing over time / new methods used
❑ organised in a network.](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-3-320.jpg)
![National
Institute
of
Science
&
Technology
[4]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Objective :
❑ Build an Ensemble classifier that can detect credit card
fraudulent transactions.
❑ Implement a classifier by use of machine learning
algorithms, such as Decision Trees, Logistic Regression,
Artificial Neural Networks and Gradient Boosting
Classifier.](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-4-320.jpg)
![National
Institute
of
Science
&
Technology
[5]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Fraud Detection System :
Machine Learning models
❑ No fixed rule
❑ Give a probability value
❑ Capture interaction
between features.
Traditional Systems
❑ Fixed threshold per rule
❑ Limited to Yes / No
❑ Fail to capture interaction
between features.
Why machine learning for fraud detection Systems ?
❑ ML model have a typical better performance.
❑ can be combined with rules.](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-5-320.jpg)
![National
Institute
of
Science
&
Technology
[6]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Software Tools :
The project is implemented by use of conda, Python, other Python Libraries &
APIs.
Steps for implementation of the project :
❑ Importing the Datasets
❑ Data Exploration
❑ Data Cleaning
❑ Data Manipulation
❑ Data modelling
❑ Ensemble Modelling](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-6-320.jpg)
![National
Institute
of
Science
&
Technology
[7]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Balancing Imbalanced Data Classes :
❑ Data Resampling - very few cases of fraud
○ Undersampling majority class. ( i.e. non-
fraud cases )
Drops data
○ Oversampling minority class. ( i.e. fraud
cases )
Creates duplicates
❑ SMOTE - Synthetic Minority Oversampling
Technique creates synthetic fraud cases by
nearest neighbours.
❑ Reframe as Anomaly / Outlier Detection](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-7-320.jpg)
![National
Institute
of
Science
&
Technology
[8]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Modelling :
❑ Supervised Learning: train model using existing fraud
labels.
❑ Unsupervised Learning: Use data to determine suspicious
behaviour without labels.
❑ Penalize Algorithms (Cost-Sensitive Training)
❑ Augment fraud detection models with Text Mining & Topic
Modelling.](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-8-320.jpg)
![National
Institute
of
Science
&
Technology
[9]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Possible Outcome :
The outcome of our work will be :
❑ An Ensemble classifier model made up of Logistic
Regression, Decision Trees, Artificial Neural Networks &
Gradient Boosting Classifier.
❑ Classifier for Fraud detection of credit cards.](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-9-320.jpg)
![National
Institute
of
Science
&
Technology
[10]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
❑ Precision: Proportion of predicted fraud
transactions that actually were true.
❑ Recall: Number of positives that were
accurately predicted.
❑ F1 score conveys the balance between the
precision and the recall.
F1 score = 2*( ( Precision * Recall ) / (
Precision + Recall ) )
Performance Analysis :](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-10-320.jpg)
![National
Institute
of
Science
&
Technology
[11]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
Reference :
[1] Masoumeh Zareapoor, Pourya Shamsolmoali,
“Application of Credit Card Fraud Detection: Based
on Bagging Ensemble Classifier”, International Conference
on Computer, Communication and
Convergence (ICCC 2015), Procedia Computer Science,
Volume 48, 2015.](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-11-320.jpg)
![National
Institute
of
Science
&
Technology
[12]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
We learned how to implement machine learning algorithms
to perform classification by Importing the Datasets, Data
Exploration, Data Manipulation, Data modelling.
Conclusion :](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-12-320.jpg)
![National
Institute
of
Science
&
Technology
[13]
B.TECH MAJOR PROJECT PRESENTATION 2020-21
Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558)
13
...Thank You…](https://image.slidesharecdn.com/majorproject-210512064832/85/Credit-Card-Fraud-Detection-13-320.jpg)
Credit Card Fraud Detection
- 1. National Institute of Science & Technology [1] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Credit Card Fraud Detection Suraj Patro CSE - 201710455 1701202348 M. Binayak Kumar Reddy CSE - 201710558 1701202192 By
- 2. National Institute of Science & Technology [2] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Contents : ❑ Introduction ❑ Objective ❑ Implementation ❑ Software Tools ❑ Possible Outcome ❑ Performance Analysis ❑ Reference ❑ Conclusion
- 3. National Institute of Science & Technology [3] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Introduction : Fraud is crime committed against property, involving the unlawful conversion of the ownership of property to one's own personal use and benefit. Some Challenges in Fraud Detection : ❑ uncommon / very few in number ❑ concealed activities ❑ changing over time / new methods used ❑ organised in a network.
- 4. National Institute of Science & Technology [4] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Objective : ❑ Build an Ensemble classifier that can detect credit card fraudulent transactions. ❑ Implement a classifier by use of machine learning algorithms, such as Decision Trees, Logistic Regression, Artificial Neural Networks and Gradient Boosting Classifier.
- 5. National Institute of Science & Technology [5] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Fraud Detection System : Machine Learning models ❑ No fixed rule ❑ Give a probability value ❑ Capture interaction between features. Traditional Systems ❑ Fixed threshold per rule ❑ Limited to Yes / No ❑ Fail to capture interaction between features. Why machine learning for fraud detection Systems ? ❑ ML model have a typical better performance. ❑ can be combined with rules.
- 6. National Institute of Science & Technology [6] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Software Tools : The project is implemented by use of conda, Python, other Python Libraries & APIs. Steps for implementation of the project : ❑ Importing the Datasets ❑ Data Exploration ❑ Data Cleaning ❑ Data Manipulation ❑ Data modelling ❑ Ensemble Modelling
- 7. National Institute of Science & Technology [7] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Balancing Imbalanced Data Classes : ❑ Data Resampling - very few cases of fraud ○ Undersampling majority class. ( i.e. non- fraud cases ) Drops data ○ Oversampling minority class. ( i.e. fraud cases ) Creates duplicates ❑ SMOTE - Synthetic Minority Oversampling Technique creates synthetic fraud cases by nearest neighbours. ❑ Reframe as Anomaly / Outlier Detection
- 8. National Institute of Science & Technology [8] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Modelling : ❑ Supervised Learning: train model using existing fraud labels. ❑ Unsupervised Learning: Use data to determine suspicious behaviour without labels. ❑ Penalize Algorithms (Cost-Sensitive Training) ❑ Augment fraud detection models with Text Mining & Topic Modelling.
- 9. National Institute of Science & Technology [9] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Possible Outcome : The outcome of our work will be : ❑ An Ensemble classifier model made up of Logistic Regression, Decision Trees, Artificial Neural Networks & Gradient Boosting Classifier. ❑ Classifier for Fraud detection of credit cards.
- 10. National Institute of Science & Technology [10] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) ❑ Precision: Proportion of predicted fraud transactions that actually were true. ❑ Recall: Number of positives that were accurately predicted. ❑ F1 score conveys the balance between the precision and the recall. F1 score = 2*( ( Precision * Recall ) / ( Precision + Recall ) ) Performance Analysis :
- 11. National Institute of Science & Technology [11] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) Reference : [1] Masoumeh Zareapoor, Pourya Shamsolmoali, “Application of Credit Card Fraud Detection: Based on Bagging Ensemble Classifier”, International Conference on Computer, Communication and Convergence (ICCC 2015), Procedia Computer Science, Volume 48, 2015.
- 12. National Institute of Science & Technology [12] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) We learned how to implement machine learning algorithms to perform classification by Importing the Datasets, Data Exploration, Data Manipulation, Data modelling. Conclusion :
- 13. National Institute of Science & Technology [13] B.TECH MAJOR PROJECT PRESENTATION 2020-21 Suraj Patro (CSE 201710455) & M. Binayak Ku. Reddy (CSE 201710558) 13 ...Thank You…
Editor's Notes
- #2: BY KHUSHBU KHAN & ISAN SAHOO
- #3: BY KHUSHBU KHAN & ISAN SAHOO
- #4: BY KHUSHBU KHAN & ISAN SAHOO
- #5: BY KHUSHBU KHAN & ISAN SAHOO
- #6: BY KHUSHBU KHAN & ISAN SAHOO
- #7: BY KHUSHBU KHAN & ISAN SAHOO
- #8: BY KHUSHBU KHAN & ISAN SAHOO
- #9: BY KHUSHBU KHAN & ISAN SAHOO
- #10: BY KHUSHBU KHAN & ISAN SAHOO
- #11: BY KHUSHBU KHAN & ISAN SAHOO
- #12: BY KHUSHBU KHAN & ISAN SAHOO
- #13: BY KHUSHBU KHAN & ISAN SAHOO