Experiment Time Series Forcasting Using Machine Learning (Case studi : Stock Value Prediction)
Abstract
The rapid development of technology has resulted in a high need for information so it is necessary to present information, one of which is the field of stocks. The use of data mining is necessary for predicting dynamic stock values. Predicting stock values by looking at certain variables using machine learning methods provides benefits for stock players. This study uses 5 attributes, namely open, low, high, close and volume on stock value. The dataset used is the shares of MNC, Hotel Sahid and XL on the Indonesia Stock Exchange during the previous 4 years. The algorithm used in this study uses machine learning including Neural Network (NN) and Support Vector Machine (SVM). The results obtained using the SVM algorithm with the smallest error value of 2,993 +/- 3,070 on MNC shares that are not that far from SVM with an error value of 3.7208 +/- 4.042. When compared with the 2 datasets of Hotel Sahid and XL shares, it is found that the NN algorithm has a smaller error value than the SVM with a long distance. So it can be concluded that the use of predictions using the Neural Network (NN) algorithm is generally better than SVM