s****l 发帖数: 10462 | 1 统计知识比较贫乏,不太懂
我有一个理论上的泊松分布,可以画一个钟罩曲线
然后我实际上观察到另外一个曲线,部分和这个理论曲线重合,两个尾巴处有些出来,
钟罩顶部比理论
的低,也就是说,象被压扁了的钟罩。我现在想评价这个实际观察到的曲线和理论上的
有多么重合,我
该使用R里面什么功能去检查?谢谢 |
h***i 发帖数: 3844 | 2 http://en.wikipedia.org/wiki/Kolmogorov%E2%80%93Smirnov_test
【在 s****l 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 统计知识比较贫乏,不太懂 : 我有一个理论上的泊松分布,可以画一个钟罩曲线 : 然后我实际上观察到另外一个曲线,部分和这个理论曲线重合,两个尾巴处有些出来, : 钟罩顶部比理论 : 的低,也就是说,象被压扁了的钟罩。我现在想评价这个实际观察到的曲线和理论上的 : 有多么重合,我 : 该使用R里面什么功能去检查?谢谢
|
a*****3 发帖数: 601 | 3 楼上言之有理。 科莫隔弱夫-西莫弱夫 test. |
s****l 发帖数: 10462 | |
l*********s 发帖数: 5409 | 5 or chisquare goodness of fit test |
s****l 发帖数: 10462 | 6 from R, I can see I need run ks.test
ks.test(x,y,alternative)
I know x is the vector that contains my actual observed numbers, and I think
I can use "two-sided" for alternative option.
What is Y? Help manual says "y: either a numeric vector of data values, or a
character string naming a cumulative distribution function or an actual
cumulative distribution function such as 'pnorm'". What should I replace '
pnorm' for Poisson distribution? Or should I used the theoretic values as y?
Thanks! |
s****l 发帖数: 10462 | 7 I used theoretic values as y in the function, it ran, but with warnings
saying that "cannot compute correct p-values with ties". What do ties mean
here? Same values within x, within y, or across x and y?
Thanks a lot! |
D******r 发帖数: 25 | 8 ppois
think
a
y?
【在 s****l 的大作中提到】![](/moin_static193/solenoid/img/up.png) : from R, I can see I need run ks.test : ks.test(x,y,alternative) : I know x is the vector that contains my actual observed numbers, and I think : I can use "two-sided" for alternative option. : What is Y? Help manual says "y: either a numeric vector of data values, or a : character string naming a cumulative distribution function or an actual : cumulative distribution function such as 'pnorm'". What should I replace ' : pnorm' for Poisson distribution? Or should I used the theoretic values as y? : Thanks!
|