The blockchain technology, especially Bitcoin (BTC) has received much attention during the past 2 years. Up to now in most of this research machine learning has been excluded. But particular attention should be paid to analyzing the blockchain with neural networks (NN), i.e. Recurrent NN (RNN) or Convolutional NN (ConvNet). The reason for this is that RNNs provide the ability for processing sequences and Bitcoin transactions are fully linked where money is transferred. This is one of the base cases in analyzing the blockchain. There are also some other problems like identifying CoinJoin transactions via generating all partitions of a set or the performance that has to be optimized because the blockchain is big (e.g. 100 GB and increasing) and nearly every byte needs to be indexed. The aim of this project is to identify such structures by leveraging machine learning algorithms, especially RNN and ConvNet, Bayes Statistics and techniques of NLP.