Publication
Application of explainable AI in Machine Learning models to identify the main determinants of Bitcoin price
| datacite.subject.fos | Ciências Sociais::Economia e Gestão | pt_PT |
| dc.contributor.advisor | Guedes, Ana | |
| dc.contributor.author | Morais, Ana Sofia Rosa | |
| dc.date.accessioned | 2023-06-26T08:30:10Z | |
| dc.date.available | 2023-12-31T01:30:39Z | |
| dc.date.issued | 2023-01-27 | |
| dc.date.submitted | 2022-12 | |
| dc.description.abstract | Since 2019, Bitcoin has become one of the most popular assets in the world. However, this decentralised cryptocurrency is typically characterised by high volatility and, in that sense, creates some concerns mainly to regulatory authorities and other decision-makers, such as governments and legislators. Furthermore, there are multiple approaches and results in the literature regarding the most relevant determinants to predict the Bitcoin price, the complexity of the Machine Learning (ML) model used to predict the Bitcoin price, and the trade-off between interpretability and the model’s performance. As a starting point, the simple model called Generalized Least Squares with Autocorrelation covariance structure (GLSAR) was found to be unrealistic to predict something as complex as the Bitcoin price. Alternatively, two more complex black box models were tested: a Long Short Term Memory neural network (LSTM) and a simple Deep Neural Network (DNN). LSTM achieved the highest 𝑅 2 score of 81.63% with DNN obtaining a 𝑅 2 score of 81.27%. Explainability techniques were applied on DNN and the results indicate that 71% of the twenty-one most significant variables are transaction-based, although future analysis can be done for occasional events. Moreover, the three most important features are the S&P500, the Bitcoin price in the previous day and how difficult it is to mine a Bitcoin block. | pt_PT |
| dc.description.abstract | Desde 2019 que o Bitcoin se tornou um dos ativos mais conhecidos no mundo. Esta criptomoeda descentralizada é tipicamente caracterizada pela elevada volatilidade e, nesse sentido, provoca algumas preocupações, sobretudo às entidades reguladoras e a outros decisores, como governos e legisladores. Além disso, há múltiplas abordagens e resultados na literatura relativamente aos fatores mais relevantes para prever o preço do Bitcoin; à complexidade do modelo de Machine Learning (ML) usado; e ao trade-off entre o nível de interpretação e o desempenho do modelo. Como ponto de partida, o modelo simples designado de Generalized Least Squares with Autocorrelation covariance structure (GLS) revelou-se irrealista para prever algo complexo como o preço do Bitcoin. Alternativamente, dois modelos black box foram testados: uma rede neural Long Short Term Memory (LSTM) e uma simples Deep Neural Network (DNN). O LSTM atingiu o melhor 𝑅 2 score de 81,63% e o DNN obteve um 𝑅 2 score de 81,27% Técnicas de explicabilidade foram aplicadas no DNN e os resultados indicaram que 71% das vinte e uma variáveis mais importantes são relacionadas com as transações, embora possam ser feitas futuras análises para acontecimentos pontuais. Moreover, the three most important features are the S&P500, the Bitcoin price in the previous day and how difficult it is to mine a Bitcoin block. Além disso, as três variáveis mais significativas são S&P500, o preço do Bitcoin no dia anterior e a dificuldade em extrair Bitcoins e hash rate. | pt_PT |
| dc.identifier.tid | 203253000 | pt_PT |
| dc.identifier.uri | http://hdl.handle.net/10400.14/41429 | |
| dc.language.iso | eng | pt_PT |
| dc.subject | Bitcoin | pt_PT |
| dc.subject | Determinants | pt_PT |
| dc.subject | LSTM | pt_PT |
| dc.subject | GLSAR | pt_PT |
| dc.subject | DNN | pt_PT |
| dc.subject | Complexity | pt_PT |
| dc.subject | Performance | pt_PT |
| dc.subject | Interpretability | pt_PT |
| dc.subject | AI | pt_PT |
| dc.subject | Decision-making | pt_PT |
| dc.subject | Determinantes | pt_PT |
| dc.subject | Complexidade | pt_PT |
| dc.subject | Desempenho | pt_PT |
| dc.subject | Interpretação | pt_PT |
| dc.subject | IA | pt_PT |
| dc.subject | Tomada de decisão | pt_PT |
| dc.title | Application of explainable AI in Machine Learning models to identify the main determinants of Bitcoin price | pt_PT |
| dc.type | master thesis | |
| dspace.entity.type | Publication | |
| rcaap.rights | openAccess | pt_PT |
| rcaap.type | masterThesis | pt_PT |
| thesis.degree.name | Mestrado em Análise de Dados para Gestão | pt_PT |