Year of Graduation
Comparison of Data Analysis Methods in Forex Forecasting
Comparison of data analysis methods in forecasting exchange rate USD/RUB was conducted in my work. Two main types of prognoses were implemented: forecast of exchange rate changes which are higher than 1% per day, and prognosis of sign of the change. In forecasting were used such data mining methods as: k-nearest neighbors algorithm, binary choice model, support vector machine, decision tree. The accuracy of prognoses was compared with the accuracy of models based on time series analysis and regularized regression. The best quality of prediction was shown by k-nearest neighbors algorithm and support vector machine and the accuracy was comparably high.