使用mle和fitdist进行R中的最大似然估计

时间:2018-04-06 21:04:47

标签: r estimation mle

我无法用最大似然估计来解决问题,你能给我一些建议吗? 我正在尝试拟合累积分布函数的参数,但我在mle()fitdist()下都有一条警告消息:

  

optim中的simpleError(par = vstart,fn = fnobj,fix.arg = fix.arg,obs = data,gr = gradient,ddistnam = ddistname,hessian = TRUE,method = meth,lower = lower,upper = upper ,...):'vmmin'中的初始值不是有限的>

函数是:1 - ((1-q)*(1-theta)/ q / theta / n)^(1 /(n-1))与未知参数theta

mle()的脚本是:

n<-as.numeric(length(price)) 

pclearing<-function(q,theta) 1-((1-q)*(1-theta)/q/theta/n)^(1/(n-1)) 

dclearing<-function(x,theta) ((1-x)*(1-theta)/x/theta/n)^((1/(n-1))-1)*((1/(n-1))*(((1-theta)/x+(1-x)*(1-theta)/x^2)/theta/n)) 

mledist(price,"clearing",start=list(theta=0.0001)) 

数据numeric名为price

0.717534108 
0.757705912 
0.598281961 
0.628852956 
0.684588176 
0.6725619 
0.639919151 
0.606366852 
0.658943911 
0.570995452 
0.576200101 
0.558362809 
0.551187468 
0.610156645 
0.620262759 
0.545224861 
0.575795856 
0.629358262 
0.69909045 
1 
0.658110157 
0.587670541 
0.639919151 
0.654850935 
0.602299141 
0.631379485 
0.685447196 
0.771576554 
0.545224861 
0.702880243 
0.655634159 
0.591207681 
0.688731683 
0.603082365 
0.698029308 
0.748863062 
0.665740273 
0.570995452 
0.709499747 
0.535219808 
0.61066195 
0.652602324 
0.729156139 
0.630368873 
0.567205659 
0.641965639 
0.689186458 
0.678120263 
0.577817079 
0.548256695 
0.714754927 
0.535219808 
0.632895402 
0.573774634 
0.59914098 
0.64426478 
0.644012127 
0.535118747 
0.703132895 
0.691258211 
0.716018191 
0.6725619 
0.71104093 
0.665740273 
0.66902476 
0.741359272 
0.719833249 
0.708084891 
0.858767054 
0.555583628 
0.674482062 
0.568468924 
0.535118747 
0.665740273 
0.548256695 
0.728941385 
0.68499242 
0.716801415 
0.568216271 
0.609903992 
0.716270844 
0.717534108 
0.717534108 
0.655634159 
0.615209702 
0.581101566 
0.568468924 

0 个答案:

没有答案