计算机应用与软件
計算機應用與軟件
계산궤응용여연건
Computer Applications and Software
2015年
10期
284-290
,共7页
股票推荐%高阶奇异分解%线性回归
股票推薦%高階奇異分解%線性迴歸
고표추천%고계기이분해%선성회귀
Stock recommendation%High-order singular value decomposition (SVD)%Linear regression
提出算法预测基金经理对股票的投资策略,为个体投资者提供投资意见。不同于仅依据股票本身信息推荐的传统算法,该算法通过高阶奇异值分解算法 HOSVD(Higher Order Singular Value Decomposition)学习基金经理的历史交易记录和投资者的个人特征因素,为投资者提供个性化推荐。除此之外,将非个性化推荐与个性化推荐进行整合,进一步提高推荐质量。对真实股票交易数据的仿真实验结果表明,用于推荐的个性化算法在准确度和收益率方面,优于传统的非个性化算法。
提齣算法預測基金經理對股票的投資策略,為箇體投資者提供投資意見。不同于僅依據股票本身信息推薦的傳統算法,該算法通過高階奇異值分解算法 HOSVD(Higher Order Singular Value Decomposition)學習基金經理的歷史交易記錄和投資者的箇人特徵因素,為投資者提供箇性化推薦。除此之外,將非箇性化推薦與箇性化推薦進行整閤,進一步提高推薦質量。對真實股票交易數據的倣真實驗結果錶明,用于推薦的箇性化算法在準確度和收益率方麵,優于傳統的非箇性化算法。
제출산법예측기금경리대고표적투자책략,위개체투자자제공투자의견。불동우부의거고표본신신식추천적전통산법,해산법통과고계기이치분해산법 HOSVD(Higher Order Singular Value Decomposition)학습기금경리적역사교역기록화투자자적개인특정인소,위투자자제공개성화추천。제차지외,장비개성화추천여개성화추천진행정합,진일보제고추천질량。대진실고표교역수거적방진실험결과표명,용우추천적개성화산법재준학도화수익솔방면,우우전통적비개성화산법。
Through predicting fund managers’investments strategy on stocks,the algorithm helps the individual investors in making rational investments decisions.Unlike traditional algorithms that solely based on stocks’information,the algorithm learns from historical transactions record of fund managers as well as the factors of personal features of investors through high-order SVD (HOSVD)algorithm to provide the personalised recommendation for investors.Besides,for further improving recommendation quality,it integrates the non-personalised and personalised recommendations.Results of simulation experiment on a real-life stock transaction dataset show that compared with traditional non-personalised algorithm,the personalised algorithm used for recommendation gains a better performance in precision and yield rate.