Today, there is a growing number of digital assets, often built on questionable technical foundations. We design and implement supervized learning models in order to explore different aspects of a cryptocurrency affecting its performance, its stability as well as its daily price fluctuation. One characteristic feature of our approach is that we aim at a holistic view that would integrate all available information: First, financial information, including market capitalization and historical daily prices. Second, features related to the underlying blockchain from blockchain explorers like network activity: blockchains handle the supply and demand of a cryptocurrency. Lastly, we integrate software development metrics based on GitHub activity by the supporting team. We set two goals. First, to classify a given cryptocurrency by its performance, where stability and price increase are the positive features. Second, to forecast daily price tendency through regression; this is of course a well-studied problem. A related third goal is to determine the most relevant features for such analysis. We compare various neural networks using most of the widely traded digital currencies (e.g. Bitcoin, Ethereum and Litecoin) in both classification and regression settings. Simple Feedforward neural networks are considered, as well as Recurrent neural networks (RNN) along with their improvements, namely Long Short-Term Memory and Gated Recurrent Units. The results of our comparative analysis indicate that RNNs provide the most promising results.

We focus on deep learning algorithms, improving upon the Weather Research and Forecasting (WRF) model, and we show that the combination of these methods produces day-ahead wind speed predictions of high accuracy, with no need for previous-day measurements. We also show that previous-day data, if available, offer a significant enhancement. Our main contribution is the design and testing of original neural networks that capture both spatial and temporal characteristics of the wind, by combining convolutional (CNN) as well as recurrent (RNN) neural networks. The input predictions are obtained by a WRF model that we appropriately parameterize; we also specify a grid adapted to each park so as to capture its topography. Training uses historical data from 5 wind farms in Greece, and the 5-month testing period includes winter months, which exhibit the highest wind speed values. Our models improve WRF accuracy on average by 19.4%, and the improvement occurs in every month; expectedly, the improvement is lowest for the park where WRF performs best. Our neural network is competitive to state-of-the-art models, achieving an average MAE of 1.75 m/s. Accuracy improves for speed values up to 20 m/s, which are important in wind energy prediction. We also develop