Cryptocurrency price prediction with artificial intelligence methods

In recent years, artificial intelligence has taken place in almost every sector, but its importance is gradually increasing. Developing technologies and increasing amount of data have made artificial intelligence inevitable in all sectors. In this study, deep learning, polynomial regression and support vector regression models from artificial intelligence methods were created, and the crypto currency bitcoin, which has entered our lives recently and is now known by a large part of the society, is selected from 2.072 data points between 01.03.2017 and 01.10.2022. trained and tested on the resulting dataset. The dataset consists of bitcoin's daily close, high, low, open, volume and market cap attributes. The models tried to predict the closing price of the target attribute, bitcoin, using the attributes in the dataset. During the training of the models, 5-fold cross validation and grid search method were used to achieve high success regardless of the data set. After the models were trained, they were tested with test data that they had not seen before and their performance values were measured. R-square, mean absolute error, and root mean square error metrics were used as performance measures. Considering the performance values of the models, it is seen that the results are close to each other, but it has been determined that the deep learning and support vector regression models are more successful than the polynomial regression model.