Cryptocurrency wallet segmentation based on machine learning techniques

OData support
Gáspár Csaba
Department of Telecommunications and Media Informatics

Ever-growing popularity of cryptocurrencies – and of Bitcoin in particular – led to a huge amount and an extensive network of transactions in the last five years of active usage. We can presume with a reason that targeting with the proper methods, this large data set laying behind cryptocurrency transactions can reveil significant, tangible information.

It is very likely that not only the topological properties of the transactions themselves but also some behavioural patterns of the users can be discovered. Since the technology behind the transactions does not allow to identify concrete persons, our personae are anonim entities who are the addresses contained in a Bitcoin wallet.

In the thesis I try to group these entities by the means of exploratory analysis and clustering of the network of transactions. The main goal is to detect and identify the abovementioned behavioural patterns. Apart from creating the clusters, analysing them individually and in relation with other segments, I also investigate their stability over time.


Please sign in to download the files of this thesis.