Skip to content

k-Nearest Neighbors (KNN) used for an Etherium Blockchain classification problem

Notifications You must be signed in to change notification settings

caozrich/Etherium_Blockchain_Phishing-_Detection_APP-KNN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Etherium Blockchain phishing detection APP using KNN

Files Logo

Ether Phishing Classifier

Contents

Description

The major blochains have been implicated in problems related to cyberattacks, scams, ponzi and pishing, the latter associated with at least 50% of cybercrime in the etherium network. this application uses a dataset of transactions in the etherium blockhain to classify by means of a KNN algorithm those that are pishing (the scope of this application is limited to the samples obtained from the dataset).

About

Allows you to retrain the algorithm in real time by changing the hyperparameters and then see the impact on the ranking performance.

New samples can be used to be classified by the previously trained model.

It also has a preview of the dataset used,(This dataset contains benign Ethereum blockchain transfers and officially reported phishing scam address transfers. The source of the dataset is the following research article

Download

This app can be downloaded from drive:

https://drive.google.com/file/d/1Gm_UUo8ijopbgOrRNGiTIbtPXzdn0_Or/view?usp=share_link

Usage: simply double-click on the Ether_Phishing_Classifier.exe file and follow the on-screen instructions.

Known issues

  • when executing, the dataset preprocessing procedure is carried out, and the KNN training, so it may take time to open depending on your system specs.
  • The file weight is 230mb due to the compression of the required data science libraries.

Libs used:

  • scikit-learn==1.2.1
  • numpy==1.24.2
  • requests==2.28.2
  • matplotlib==3.7.0
  • pandas==1.5.3
  • seaborn==0.12.2
  • PyQt5==5.15.4

Support:

If you have any questions or problems with the app, please contact the author at libreros00@gmail.com