3
1

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 3 years have passed since last update.

2.4(標準) 期待値の定義

Last updated at Posted at 2021-07-09

方針

いずれもtについての一変数関数なので,前半は平方完成,後半は微分して0と置くことで最小値を計算する.定積分の微分について,以下であることを用いる.

\frac{d}{dx}\int_a^xf(t)dt=f(x)

また,「分布f_Xに従う確率変数Xについての期待値」を,次のように書くことにする.

\mathbb{E}_{X\sim f_X}

答案

\begin{align}
\mathbb{E}_{X\sim f_X}[(X-t)^2]&= \mathbb{E}_{X\sim f_X}[X^2]-2t\mathbb{E}_{X\sim f_X}[X]+t^2\\
&= (t-\mathbb{E}_{X\sim f_X}[X])^2+\mathbb{E}_{X\sim f_X}[X^2]-(\mathbb{E}_{X\sim f_X}[X])^2
\end{align}

より,

\arg\min_t\mathbb{E}_{X\sim f_X}[(X-t)^2]=\mathbb{E}_{X\sim f_X}[X]

となる.次に,

\begin{align}
\mathbb{E}_{X\sim f_X}[|X-t|]&= \int_{–\infty}^\infty|x-t|f_X(x)dx\\
&= \int_t^\infty(x-t)f_X(x)dx-\int_{-\infty}^t(x-t)f_X(x)dx\\
&= \int_t^\infty xf_X(x)dx-t\int_t^\infty f_X(x)dx-\int_{-\infty}^txf_X(x)dx+t\int_{-\infty}^tf_X(x)dx\\
&= (\int_t^\infty xf_X(x)dx-\int_{-\infty}^txf_X(x)dx)-t(\int_t^\infty f_X(x)dx-\int_{-\infty}^tf_X(x)dx)\\
&= (\mathbb{E}_{X\sim f_X}[X]-2\int_{-\infty}^txf_X(x)dx)-t(1-2\int_{-\infty}^tf_X(x)dx)
\end{align}

と変形できる.これをtについて微分して0と置く.

\begin{align}
\frac{\partial\mathbb{E}_{X\sim f_X}[|X-t|]}{\partial t}&= -2tf_X(t)-(1-2\int_{-\infty}^tf_X(x)dx)+2tf_X(t)=0\\
\Rightarrow \int_{-\infty}^t f_X(x)dx&= \frac{1}{2}
\end{align}

ゆえ,E[|X-t|]を最小化するtは,P[X<1/2]=tとなるような点,すなわちXの分布の中央値である.

参考文献

3
1
3

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
3
1

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?