Skip to content

Decision Trees, Ensembling, Adaboost and Random Forest on Bitcoin Heist Ransomware Address Dataset

Notifications You must be signed in to change notification settings

havelhakimi/Bitcoinheistransomware

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 

Repository files navigation

Machine Learning Models for Bitcoin Heist Ransomware Address Dataset

This is a solution notebook to an assignment question given in a Data Mining graduate course. Each code block is accompanied by relevant analysis wherever required.
Dataset link : https://archive.ics.uci.edu/ml/datasets/BitcoinHeistRansomwareAddressDataset
Broadly, the following steps have been performed in this solution notebook:

  • A custom designed train-test split method which splits the data into training, validation and test set (70:15:15).
  • Decision Tree trained using both the Gini index and the Entropy by changing the max-depth as [4,8,10,15,20].
    • The splitting criteria (gini/entropy) selected is the one which gives better accuracy on test set with the chosen depth.
  • Ensembling method is a method to combine multiple not-so-good models to get a better performing model.
    • Created 100 different decision stumps (max-depth 3). For each stump trained it on a randomly selected 50% of the training data i.e. selelct data for each stump separately
    • Finally predicted the test samples's labels by taking a majority vote of the output of the stumps.
  • Used the sklearn Adaboost algorithm on the above dataset and reported the testing accuracy.
    • Decision tree is used as the base estimator and with a number of estimators as [4,8,10,15,20].
    • Further compared the Random Forest and Adaboost results.
These above assumptions and the flow of work is according to the questions asked in assignment.

About

Decision Trees, Ensembling, Adaboost and Random Forest on Bitcoin Heist Ransomware Address Dataset

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published