Fateme Zahra Darzi - MSc student; Alzahra university;
Roshanak Alimohammadi - Associate professor; Alzahra university;
Seyed Bagher Mirashrafi - Associate professor; Mazandaran university;

The gold price forecast is similar to the forecasting of other financial time series. In this paper, we make forecasts using data mining methods. The data set used in this article is the gold price data set and it s from 1985 to 2019 and it has 8569 rows. In this paper, we try to achieve better forecasting by defining more useful features on the dataset and eliminating some inefficient features. The five features of the high minus low percent(HML), price changes, random oscillator, %D and position, are defined on the dataset and their effects on model are examined. These investigations are performed by three data mining algorithms i.e. SVR, Random Forest Regression and KNN. Finally, among all the algorithms we tried, we selected an algorithm that has lower RMSE and calls it the best algorithm among the algorithms we used for model prediction.

Financial Forecasting, Data Mining, SVR, Random Forest Regression