CAJ | 학술논문

采用神经网络 Elman网络模型对汇率进行预测.处理过程对样本序列进行了分类，并对训练与测试样本进行了残差分析.预测结果表明：该方法对汇率涨落方向的预测准确度达到74.54%，对汇率预测值与实际值之间的偏差略为偏大的情况，并分析了产生误差的原因.
채용신경망락 Elman망락모형대회솔진행예측.처리과정대양본서렬진행료분류，병대훈련여측시양본진행료잔차분석.예측결과표명：해방법대회솔창락방향적예측준학도체도74.54%，대회솔예측치여실제치지간적편차략위편대적정황，병분석료산생오차적원인.
This paper predicts the Exchange Rate by employing Neural Networks(NN)models.We classified the sample sequence,and then made residual analysis on the training and test samples.The predicted results show that the method has a 74.54% accuracy to predict the fluctuation of Exchange Rate.Furthermore,we analyzed the phenomenon of a slightly big devi-ation between the Exchange Rate’s predicted values and actual value.