数学的实践与认识
數學的實踐與認識
수학적실천여인식
MATHEMATICS IN PRACTICE AND THEORY
2009年
20期
31-34
,共4页
最优分区间%单位相似度向量%预测
最優分區間%單位相似度嚮量%預測
최우분구간%단위상사도향량%예측
optimization partition%unit similarity vector%similarity agorithm%prediction
通过定义了一种基于数据最优分区间相似度算法,利用学习样本得单位相似度向量,并得各维数据的最优分区间,利用最优分区间得预测样本与学习样本的单位相似度向量,从而得预测样本的预测值.通过实例表明,算法所预测的结果相对误差可达百分位,并且本算法能应用到其它数据处理中,具有较广泛的通用性.
通過定義瞭一種基于數據最優分區間相似度算法,利用學習樣本得單位相似度嚮量,併得各維數據的最優分區間,利用最優分區間得預測樣本與學習樣本的單位相似度嚮量,從而得預測樣本的預測值.通過實例錶明,算法所預測的結果相對誤差可達百分位,併且本算法能應用到其它數據處理中,具有較廣汎的通用性.
통과정의료일충기우수거최우분구간상사도산법,이용학습양본득단위상사도향량,병득각유수거적최우분구간,이용최우분구간득예측양본여학습양본적단위상사도향량,종이득예측양본적예측치.통과실례표명,산법소예측적결과상대오차가체백분위,병차본산법능응용도기타수거처리중,구유교엄범적통용성.
A algorithm of similarity on the data optimization partition is definited, unit similarity vector and data optimization partition are got through study samples. Using optimization partition, the unit similarity vector between forecast samples and study samples is computated, and prediction value of prediction sample is gained. The examples show that the Relative error of results got by the algorithm has the one-thousandth, and the algorithm can be applied on other data and has widespread versatility.
