24小时热门版块排行榜    

查看: 255  |  回复: 0

mike880102

金虫 (著名写手)

[求助] 求解一个关于统计的问题

在一段文字中看到,先使用第一个函数求解参数(This function treats each group as an individual entity,even though in reality, it belongs to a population with similar characteristics.);如果误差较大,则继续用第二个函数求解(This step considers the individual groups as belonging to a sum of random- and fixed-effects),从而达到减小误差的目的。
这里为什么将这些groups 看作是同属一个 population之后,就可以提高accuracy?应涉及到独立性的问题了吗?还是相关性的问题?

英文原文如下:
HydroMe has two steps for estimating parameters for grouped data. The first step makes a first approximation of the parameter for each group using the nlsList function. This function treats each group as an individual entity, even though in reality, it belongs to a population with similar characteristics. The function can sometimes produce estimates with large standard errors for groups with high scatter or outlying observations. Estimates with large standard errors are generally the results of low estimation accuracy (Gourdin and Boumahrat, 2002). If convergence errors are not reported for such cases, the low estimation accuracy may not be noticed. This explains the occurrence of low accuracy in existing computer programs where the estimation process treats each sample as an independent individual. The Gauss–Newton algorithm in HydroMe always reports convergence errors if the estimation accuracy exceeds a preset tolerance. These convergence errors should not raise too much alarm other than a warning to go back and check the reasons behind the scatter. Otherwise, the package takes care of such convergence errors in the next step. The focus in the second step is to improve the estimation accuracy. This step considers the individual groups as belonging to a sum of random- and fixed-effects gives the estimated hydraulic parameters. The second step of the hydraulic parameter estimation is achieved using the nlme function.
回复此楼
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

智能机器人

Robot (super robot)

我们都爱小木虫

找到一些相关的精华帖子,希望有用哦~

科研从小木虫开始,人人为我,我为人人
相关版块跳转 我要订阅楼主 mike880102 的主题更新
信息提示
请填处理意见