In this article, we explore the complexity of cryptocurrencies by focusing on predicting the price of Bitcoin, a key issue for investors. Using correlation analysis between Bitcoin price and various indicators, we developed four prediction models. The Random Forest model proved to be relevant, as it obtained a prediction of the Bitcoin price to within 18 dollars. As part of the challenge proposed by the EGC 2024 conference, this study also revealed unexpected correlations between anomalies between transactions of different actors and variations in the price of Bitcoin. Analysis of the transaction network also revealed surprising trends, such as the increase in the clustering coefficient during periods of Bitcoin growth and the decrease in the number of communities during its rises.