Try   HackMD

A.3.3 Multinomial Distribution

定義

延續

A.3.1,如果我們將 Bernoulli distribution (只有兩個 states,"success"(
1
)和 "failure"(
0
))generalize 到多個 states:

也就是說一個 random event 的 outcome 是

K 個 mutually exclusive 且 exhaustive 的 states 中的其中一個。

  • mutually exclusive:任兩個相異的 states 相交
    =ϕ
    ,也就是說不會有兩個 states 同時發生
  • exhaustive:所有的
    K
    個 states 聯集
    =
    sample space,也就是所有可能的 outcomes 都被涵蓋在這
    K
    個 states 中。

那麼若這

K 個 states 每個發生的機率都令為
pi
,則滿足:

i=1Kpi=1

假設有

N 個這樣的 trials,其中 outcome 是
i
的次數令為
Ni
,且:

i=1KNi=N

N1,N2,...,NK 的 joint distribution 是 multinomial 的:

Image Not Showing Possible Reasons
  • The image was uploaded to a note which you don't have access to
  • The note which the image was originally uploaded to has been deleted
Learn More →

特例

A.38 的特例是當
N=1
時:

我們只有一個 trial(只執行這個 random experiment 一次),則所有的

Ni 都是
0/1
indicator

因為只有一個 trial,所以 outcome 只有一個,因此所有的 states 裡面只有一個會發生,因此只有一個

Ni 會是
1
,其他都是
0

滿足:

Image Not Showing Possible Reasons
  • The image was uploaded to a note which you don't have access to
  • The note which the image was originally uploaded to has been deleted
Learn More →