现代电子技术
現代電子技術
현대전자기술
MODERN ELECTRONICS TECHNIQUE
2015年
9期
38-42,46
,共6页
复杂产品%支持向量回归机%小样本%费用估算
複雜產品%支持嚮量迴歸機%小樣本%費用估算
복잡산품%지지향량회귀궤%소양본%비용고산
complex product%support vector regression%small sample%cost estimation
传统的费用估算算法需要大量的样本数据来保证其估算的准确性,但在实际应用中,由于样本数据的有限性,其准确性无法得到保证,针对这种情况提出使用基于统计学习理论的支持向量回归机(SVR)进行费用估算,并通过具体实例详细阐述基于SVR的费用估算具体步骤,包括数据预处理、基于SVR的训练、估算和后处理过程,通过与神经网络方法相比,实验结果验证了SVR在小样本情况下具有更好的估算精度。最后实现了基于SVR的复杂产品费用估算方法,并集成于复杂产品费用估算系统。
傳統的費用估算算法需要大量的樣本數據來保證其估算的準確性,但在實際應用中,由于樣本數據的有限性,其準確性無法得到保證,針對這種情況提齣使用基于統計學習理論的支持嚮量迴歸機(SVR)進行費用估算,併通過具體實例詳細闡述基于SVR的費用估算具體步驟,包括數據預處理、基于SVR的訓練、估算和後處理過程,通過與神經網絡方法相比,實驗結果驗證瞭SVR在小樣本情況下具有更好的估算精度。最後實現瞭基于SVR的複雜產品費用估算方法,併集成于複雜產品費用估算繫統。
전통적비용고산산법수요대량적양본수거래보증기고산적준학성,단재실제응용중,유우양본수거적유한성,기준학성무법득도보증,침대저충정황제출사용기우통계학습이론적지지향량회귀궤(SVR)진행비용고산,병통과구체실례상세천술기우SVR적비용고산구체보취,포괄수거예처리、기우SVR적훈련、고산화후처리과정,통과여신경망락방법상비,실험결과험증료SVR재소양본정황하구유경호적고산정도。최후실현료기우SVR적복잡산품비용고산방법,병집성우복잡산품비용고산계통。
Since plenty of sample data is required to ensure the accuracy of traditional cost estimation algorithm,and it is hard to ensure the accuracy of estimation due to the limitation of sample data in practical application,the support vector regres?sion(SVR)based on statistical learning theory is proposed to make cost estimation. The specific steps of cost estimation is de?scribed in detail based on SVR,including data preprocessing,training based on SVR,estimation and post?processing. The ex?periment result verifies that the estimation accuracy based on SVR in small sample data is better than the method of neural net?work. Finally,the method of complex product cost estimation based on SVR is implemented,and is integrated in the system of complex product cost estimation.