CAJ | 학술논문

本文借助校4A教学平台，在数据挖掘教学过程中，建立实验模型。以上证指数2009年9月21日至2014年4月24日收盘价以及技术指标的历史数据为依据，首先利用RBF神经网络对上证指数时间周期为5 min的每日收盘价的最低价进行预测建模和比较试验，经过与上证收盘指数的真实值的比较发现，神经网络在预测长期股票趋势时具有较好的效果，所得的归一化之后的误差平方和比较小。其次得到上证指数收盘价与技术指标都显著相关，对技术指标进行pearson相关性分析，得到三个技术指标RSI、DMI、FSL不互相显著相关，在此基础上，RSI、DMI、FSL作为描述属性，上证指数收盘价作为类别属性构造决策树，评估结果给出了具体的类别方案。实验教学效果较好。
본문차조교4A교학평태，재수거알굴교학과정중，건립실험모형。이상증지수2009년9월21일지2014년4월24일수반개이급기술지표적역사수거위의거，수선이용RBF신경망락대상증지수시간주기위5 min적매일수반개적최저개진행예측건모화비교시험，경과여상증수반지수적진실치적비교발현，신경망락재예측장기고표추세시구유교호적효과，소득적귀일화지후적오차평방화비교소。기차득도상증지수수반개여기술지표도현저상관，대기술지표진행pearson상관성분석，득도삼개기술지표RSI、DMI、FSL불호상현저상관，재차기출상，RSI、DMI、FSL작위묘술속성，상증지수수반개작위유별속성구조결책수，평고결과급출료구체적유별방안。실험교학효과교호。
Based on 4A Teaching platform and in the teaching of data mining, the following research was done. Based on the closing price and technical indicators from September 21,2009 to April 24,2014 of Shanghai securities composite indexes. Firstly, this paper made the prediction and comparison to lowest price of the daily closing price of Shanghai securities composite indexes at minutes by using RBF neural network. By comparing the prediction results with actual values of Shanghai securities closing price, we find that the neutral network model has shown a satisfying performance in predicting the long-term stock trends. Based on the normalized data, the sum of squared errors is relatively small. Secondly, we get the Shang securities closing price and technical indicators are significantly related. We made analysis to technical indicators and get that three technical indicators RSI, DMI, FSL are not significantly related each other. On this basis, we construct a decision tree with RSI, DMI, FSL as described attribute and the Shanghai securities index closing price as the class attribute. Model results suggest that the maximum of information gain is RSI. And we can get the specific category programs.