Un système basé sur l’apprentissage automatique pour la prédiction de la valeur des cryptomonnaies

Abstract

iii ABSTRACT Cryptocurrencies have indeed experienced growing interest and popularity in recent years as digital assets independent of central banking institutions. In our re- search and development of our model, we have primarily focused our efforts on pre- dicting the price of Bitcoin, the flagship and most widely recognized cryptocurrency. We have tried several models trained on different datasets and economic and finan- cial indicators, and the best results were obtained using an MLP (Multi-Layer Percep- tron) model trained on a database containing various variations of the RSI (Relative Strength Index) economic indicator calculated over four different periods.