Summary

In conclusion, we have identified and discussed one of the key threats in the cryptocurrency space, highlighting the need for effective transaction monitoring and identification. To this end, we have undertaken a machine learning exercise at the Ethereum address level, where we have leveraged Etherscan to complete our dataset.

We have evaluated and compared various machine learning models, optimizing their performance through grid search hyperparameter tuning and cross-validation. By undertaking this project, we have dived into a subject matter where forensics professionals are active and remains a current news topic.

Blockchain forensics is one of the more innovative areas in data science applications, as models need to scale and keep evolving in order to adapt, to be able to spot new types of fraud and scams.

In the next chapter, we will dive into predicting prices.