本文選題：混合再生散度模型　切入點(diǎn)：EM算法　出處：《統(tǒng)計(jì)與信息論壇》2017年08期 　論文類型：期刊論文

【摘要】：隨著社會信息化的發(fā)展數(shù)據(jù)的種類越來越多樣化,在實(shí)際的數(shù)據(jù)分析中相對于同質(zhì)總體,異質(zhì)總體更具有普遍性,所以混合回歸模型是最重要的統(tǒng)計(jì)數(shù)據(jù)分析工具之一。再生散度模型是一種比指數(shù)族分布更加廣泛的分布,適用性更強(qiáng)�；诖�,提出混合再生散度模型,對方位參數(shù)進(jìn)行建模,并通過EM算法研究該模型參數(shù)的極大似然估計(jì)。同時,通過隨機(jī)模擬和實(shí)例研究說明該模型和方法是有效和有用的。
[Abstract]:With the development of social information data types are more and more diversified, in the analysis of the actual data with respect to the homogeneous population, overall heterogeneity is more universal, so mixed regression model is one of the most important tools for statistical analysis of data. The reproductive dispersion model is a number of family distribution ratio refers to the broader, more applicable. Based on the proposed hybrid reproductive dispersion model, modeling the range of parameters, the maximum likelihood and the EM algorithm of the parameter estimation of the model. At the same time, through random simulation and case study show that the model and method is effective and useful.

【作者單位】： 昆明理工大學(xué)理學(xué)院;
【基金】：國家自然科學(xué)基金項(xiàng)目《復(fù)雜數(shù)據(jù)下聯(lián)合均值與方差模型的統(tǒng)計(jì)推斷》(11261025) 國家自然科學(xué)基金項(xiàng)目《復(fù)雜空間點(diǎn)過程數(shù)據(jù)的統(tǒng)計(jì)推斷》(11126309)
【分類號】：O212.1
，

本文編號：1557208