- Tytuł:
-
Porównanie podejścia aproksymującego i klasyfikującego w prognozowaniu kursów wybranych akcji na GPW w Warszawie S.A. z użyciem jednokierunkowych sieci neuronowych
Forecasting Stock Prices Using Feed-Forward Neural Network - a Comparison of Approximation and Classification Approaches - Autorzy:
- Kasznia, Anna
- Powiązania:
- https://bibliotekanauki.pl/articles/589117.pdf
- Data publikacji:
- 2013
- Wydawca:
- Uniwersytet Ekonomiczny w Katowicach
- Tematy:
-
Giełda papierów wartościowych
Kurs akcji
Prognozowanie
Sieci neuronowe
Szeregi czasowe
Forecasting
Neural networks
Share price
Stock market
Time-series - Opis:
- In this paper two approaches to financial time series forecasting using neural networks were compared. First one, the function approximation approach, in which neural networks are trained to forecast the exact one day ahead value of stock price. And the second one, classification approach, in which the output variable is the direction of future stock price movements. The aim of this work was to check if using the classification models can lead to better results in terms of direction of change forecasting and profits generated by their forecasts. This research was conducted on the basis of the time series of daily closing stock prices for three companies listed on the Warsaw Stock Exchange. Simulations show that some of the approximating models achieved satisfactory results in terms of the directional symmetry measure, although the best results for each of the analyzed company have been achieved for classification models.
- Źródło:
-
Studia Ekonomiczne; 2013, 146; 59-67
2083-8611 - Pojawia się w:
- Studia Ekonomiczne
- Dostawca treści:
- Biblioteka Nauki
Artykuł