The project presentation focuses on identifying charge card fraud using various machine learning algorithms, highlighting the rise in card misrepresentation due to online transactions. It proposes a system using a random forest algorithm, among others, to predict fraudulent activities by analyzing transaction data. Future enhancements aim to improve the accuracy of these algorithms for better outcomes.

1. A Project Presentation
on
CHARGE CARD MISREPRESENTATION
IDENTIFICATION
1
K.SRINIVAS 3VC18SCS05
Under the Guidanceof
Dr. SAPNA B KULKARNI
AssistantProfessor,Dept. of CSE
RAO BAHADUR Y MAHABALESWARAPPA ENGINEERING COLLEGE
Departmentof ComputerScience&Engineering

2. Agenda
2
Introduction
SystemAnalysis
SystemRequirements
Implementation
Conclusion
FutureEnhancement
Snap shots

3. INTRODUCTION
• Master card are broadly stratified due to the increase of online
merchandising and the upswing of many-sided keen widgets.
• Master card is letting us to make net transactions in a simple manner so
many of them are using it.
• As per recent research the card misrepresentation and faking of the cards
been done in increasing manner, which is resulting into affliction of
money ever year.
• Evaluations say that afflictions are drastically rising at paired digit rates
by 2020.
3

4. Continued.,
• Charge card extortion occasions happen as often as possible at that
point end in enormous money related misfortunes.
• Hoodlums can utilize a few innovations like Trojan or Phishing to
take the information of others' cards.
• In this way, a productive extortion discovery technique is
indispensable.
• Since it can distinguish a misrepresentation in time when a criminal
uses a taken card to devour.
• One strategy is to make full utilization of the recorded exchange
information including typical exchanges and misrepresentation.

5. • One strategy is to make full utilization of the recorded exchange
information including typical exchanges and misrepresentation.
• We can use Machine learning algorithms to detect whether the
transaction is genuine or not.
• The data used in our investigations originate from an online business
organization.
Continued.,

6. SYSTEM ANALYSIS
EXISTING SYSTEM
• When the card is not present the trade-off is increasingly mainstream,
peculiarly almost Master card activities are performed by many online
portals.
• Master card is letting us to make net transactions in a simple manner so many
of them are using it.
• As per recent research the card misrepresentation and faking of the cards
been done in increasing manner, which is resulting into affliction of money
ever year.
• The major disadvantage of the existing system is the fundamental
inconvenience of the current framework is the discovery happens simply after
gets a composed grievance.

7. Continued.,
• In proposed framework, we use abuse technique which can request
that the PC see if it's charge card misrepresentation or not.
• Here we utilized Random Forest calculation that investigates and
predicts the extortion and non-misrepresentation exchanges.
• We use one of the ensemble method in this proposed work which is
the aggregation of several tree predictors specified every tree rely on
a arbitrary autonomous set of data .
• Every tree under the forest has equal allocation.

8. ADVANTAGES OF PROPSED SYSTEM
• Performance is acceptable.
• Reduces the time required to anticipate the yield.
• Used for ongoing expectations of misrepresentation exchanges.
• The intent of our project is to find out the Charge card Fraud using
the machine learning algorithms.
• We are using various machine learning algorithms to find the
fraudulent transactions.
• The algorithms used are random forest, SVM ,KNN, Naive bayes to
get the best accuracy, precision and recall score to determine the
fault transactions from the input data set.

9. SYSTEM REQUIREMENTS
HARDWARE REQUIREMENTS:
• 10 GB HDD (min)
• 128 MB RAM (min)
• Pentium P4 Processor 2.8 Ghz (min)
SOFTWARE REQUIREMENTS:
• Python 3.7 or higher IDE
• MySQL
• GUI

10. IMPLEMENTATION
• SYSTEM ARCHITECTURE

11. • DATA FLOW DIAGRAM

12. STEPS INVOLVED IN PROCESS:
• USER LOGIN: The user log in the project using the My-SQL for the
security purpose. First user registers and then registered data is
stored in the My-SQL and which helps to store user credentials and
login the project with the user name and password.
• PREPROCESSING :The dataset is used to read in the machine and
its transformed from raw data into clean dataset. Hence the machine
can understand the parameters and the different data types in the
dataset.

13. • FEATURE EXTRACTION :
In our project, we use the dimensionality reduction process by
which an initial set of raw facts is compressed to more viable
clusters for processing. The change of the original data is generated
in the data set with a low number of variables.
• The dataset is having the principle parameters of Time, Amount and
Class. Utilize these parameters we can anticipate the deceitful and
non-fake exchanges of the charge card.

14. • PREDICTION:
We have used RF, SVM, KNN, NAIVEBAYES algorithms to
analyze and predict the fraud and non-fraud/valid transactions.
• RANDOM FOREST ALGORITHM:
A supervised algorithm which is an ensemble of decision
trees. Here, we’ve collection of decision trees. The decision
trees are the building blocks of the RF model. We use this
approach to predict the Master card fraud through ML
technology.

15. • SUPPORT VECTOR MACHINE:
A supervised category set of rules, that plot a line that divides
distinctive categories of your data. It is taken into consideration to
be a classification approach, but may be employed in both kinds of
classification and regression problems. It can easily deal with
multiple non-stop and specific variables.
• NAIVE BAYES ALGORITHM:
Naive Bayes is a supervised classification algorithm method which
relies on Bayes theorem. A classification approach with an
assumption of independence among predictors. This classifier
assumes that the presence of a particular feature in a class is
unrelated to the presence of any other feature.

16. • K-NEAREST NEIGHBOR’S ALGORITHM:
KNN is a supervised algorithm that considers extraordinary
centroids and makes use of a commonly Euclidean characteristic to
compare distance.
• K stands for variety of the nearest neighbouring points. We use
KNN set of rules for the prediction of the MasterCard fraud through
ML technology.

17. FLOW CHART:

18. • If we compare the all four algorithms we have the following results:
• As per the project output the comparison of the four algorithms with
their accuracy, precision and recall score is in the percentage as
shown below in the table.
Algorithm Accuracy Precision Recall
Random Forest 0.9995 0.9238 0.8016
SVM 1.000 1.0000 1.0000
Naïve Bayes 1.000 1.000 1.000
KNN 0.9993 0.7777 0.0614

19. SNAP SHOTS

21. FUTURE ENHANCEMENT
• In future work we concentrate on improving the accuracy of random
forest algorithm and its calculations to get the best outcomes.
• In this way, we likewise attempt to make some improvement for this
calculation.

22. CONCLUSION
• This project has inspected the exhibition of Random Forest, SVM,
Naïve Bayes and KNN algorithms.
• A genuine B2C dataset on charge card exchanges is utilized in our
examination.
• The calculation of arbitrary timberland itself ought to be improved.
In this way, we likewise attempt to make some improvement for this
calculation.
• we need to concentrate on improving the accuracy of random forest
algorithm and its calculations to get the best outcomes.