MSc Computer Science project pertaining to the detection of illicit accounts over the Ethereum network based on the respective accounts transaction history. Using the XGBoost classification model, we carry out binary classification on a dataset containing a near balanced datset of normal and illicit accounts.
The dataset is present in \Account_Stats titled Complete.csv with a total of 4681 accounts (2179 illicitly and 2502 normal accounts). Each account has 42 extracted features.
The \Illicit_Accounts directory contains the respective code to obtain all illicit accounts from the EtherscamDB repository.