Bitcoin Transactions Analysis

We studied the bitcoin transactions from 03/01/2009 to 09/09/2016. Using neo4j, we did a clustering of the bitcoin addresses, and constructed the bitcoin network user graph (a graph database).

Prerequisites

Software

• Python 3.7
• Neo4j Community Edition 3.5.6
• The machine we used had 64 RAM, 19 processors

Python Modules

• jupyter notebook
• numpy
• pandas
• py2neo

Data to analyse

• Bitcoins Transactions : JSON format
• The labels of the BTC Addresses (optional)

The project Folder "Code_StageM2_BITUNAM"

Has 4 sub-folders : source, run, neo4j_scripts and results.

notebook 'Execute_the_scripts.ipynb' :

• describes how to run the 4 scripts contained in the run folder
• does the clustering graph and export the clusters
• explains how to construct the Bitcoin network graph using Neo4j.

Folder source

Contains all the functions used for this project.

• functions_check_in_data.py
• functions_write_save_load_data.py
• functions_extract_transform_data.py
• functions_for_the_clustering.py
• functions_for_bitcoin_graph

Folder run

Contains 4 scripts:

• create_data_clustering_h1.py : creates the dictionary of addresses, converts the bitcoin addresses to int, creates the data
  for the clustering graph with h1.
• add_heuristic_change_address.py : used to add the change addresses to the clustering data with h1.
• build_clusters.py : Used to create the dictionary of the clusters.
• create_data_bitcoin_graph.py : used to create the nodes and links for the bitcoin graph construction.

Folder neo4j_scripts

Contains 3 Neo4j scripts, that must be executed on a SSH.

• neo4j_script_clustering_graph.sh: used to build the clustering graph on Neo4j
• neo4j_script_export_partitions.sh: used to export partition to the working directory
• neo4j_script_bitcoin_graph.sh: used to build the Bitcoin Graph in Neo4j

The folder contains also 3 scripts (.bat files) which can be used in a windows terminal.

Folder results

Contains 4 Notebooks:

• create_dictionary_of_identities.ipynb: creates the dictionary of identities (labels)
• results_list_of_lists.ipynb: Results obtained with the list of lists of addresses
• results_clusters.ipynb: Statistics of the clusters
• neo4j_queries.ipynb: Some statistics obtained from the Bitcoin user graph.

Author:

Boubacar SOW

Contact:

2s.boubacar@gmail.com