https://github.com/daniel-elston/data-mining-fraud-outlier-detection
Data mining techniques used to identify outliers and potential fraudulent activity in a stocks dataset.
https://github.com/daniel-elston/data-mining-fraud-outlier-detection
data-mining data-science fraud-detection machine-learning
Last synced: over 1 year ago
JSON representation
Data mining techniques used to identify outliers and potential fraudulent activity in a stocks dataset.
- Host: GitHub
- URL: https://github.com/daniel-elston/data-mining-fraud-outlier-detection
- Owner: Daniel-Elston
- Created: 2022-03-08T14:07:30.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2022-06-11T21:30:59.000Z (about 4 years ago)
- Last Synced: 2025-02-09T16:36:47.146Z (over 1 year ago)
- Topics: data-mining, data-science, fraud-detection, machine-learning
- Language: Jupyter Notebook
- Homepage:
- Size: 236 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
Data Mining - Fraud Outlier Detection
## Table of contents
- [Status and Details](#status-and-details)
- [Technology](#technology)
- [Introduction](#introduction)
- [Project Description](#project-description)
- [Objectives](#objectives)
- [Data Science Methodology](#data-science-methodology)
- [Problem Formulation](#problem-formulation)
- [Conclusions](#conclusions)
- [Contributing Members and Contacts](#contributing-members-and-contacts)
## Status and Details
- **Project Status**: [Completed]
- **Date Coded**: 07/10/21
## Technology
- **Language**: Python 3.6.9
- **Libraries**: numpy, matplotlib, pandas, sklearn
- **Set up File**: N/A
## Introduction
The purpose of this project is to identify potential fraudulent activity. Such analysis is important in the financial sector.
### Project Description
The dataset used for this analysis is a stock dataset with over 2500 observations. The data denotes the percentage of changes in the daily closing price of stocks for Microsoft, Ford and Bank of America.
A one-class SVM is trained on the dataset to identify outliers, potentially preventing fraudulent activity. The results are plotted on a 3D plot with a colour coded system to denote outlier score.
### Objectives
- Train one-class SVM to return outlier labels
- Plot results to visualise potential outliers
## Data Science Methodology
The below subsections outline the standard methodology of data scientists.
### Problem Formulation
Market manipulation can causes harm to investors in such companies. By reducing fraudulent stock activity, investors can be protected.
## Conclusions
The resultant plots for this project clearly show outliers in a dark red. The closer to thje center of the plot, the closer to blue the colour and also the less likely the observation is to be fraudulent.
## Contributing Members and Contacts
**Team Lead: [Daniel Elston](https://github.com/Daniel-Elston)**
|Name | GitHub Handles |
|---------|-----------------|
| Daniel Elston | [Git DE](https://github.com/Daniel-Elston) |
Please feel free to contact me if you have any questions, require any further information or wish to contribute.
Email 1: delstonds@outlook.com
Email 2: ec21024@qmul.ac.uk