PENERAPAN TEKNIK JARINGAN SYARAF TIRUAN ALGORITMA BACKPROPAGATION UNTUK PERAMALAN HARGA SAHAM

ADYANTO, PUTRA CHRISTIAN (2014) PENERAPAN TEKNIK JARINGAN SYARAF TIRUAN ALGORITMA BACKPROPAGATION UNTUK PERAMALAN HARGA SAHAM.

Preview

PDF - Published Version Download (4Kb) | Preview

Abstract

Forecasting stock is one of the important inputs for the investor in making an investment decision. In the forecasting process can be realized that the results often inaccurate predictions, but why prediction is still necessary? The answer is all organizations operate in an environment that contains elements of uncertainty, but the decision must still be taken which will affect the future of the organization. A scientific prediction for future more meaningful than simply relying on intuition. The mathematical model of a problem is sometimes difficult to be made and that there was just the data, as well as on changes in stock prices that is difficult to resolve and unpredictable. Neural network does not require a mathematical model but using data from the problem to be solved. Neural network using data provided and processing data in the training process. So it is very appropriate neural network to solve the problem of forecasting, especially forecasting stock prices. The data of stock prices can be enforced in time series. If we have the data in a certain period, the share price of the stock in the next period can be predicted (time spent can be hourly, daily, weekly, monthly or yearly). So forth is done repeatedly until N iterations weekdays. For good results the prediction of neural network must be likened to an input that represents some aspect of the stock price. Then a weighting principle to minimize the prediction error one step ahead. With the final weights do an action to minimize the total error for the next iteration.

Actions (login required)