Обсуждение участника:Riabenko

Материал из MachineLearning.

(Различия между версиями)
Перейти к: навигация, поиск
Строка 1: Строка 1:
-
{{MediaWiki:NewUserMessage|Riabenko}}
+
[[http://isi.cbs.nl/glossary/index.htm Глоссарий статистических терминов ISI]]
 +
 
 +
 
<tex>\mathbb{E}\left[1\left\{K_j^{(b)}\leq\alpha\gamma\right\}\right] \leq \mathbb{P}\left[\left.P_j^{(b)}\leq\alpha\gamma\right|S\subseteq \tilde{S}^{(b)}\right].</tex>
<tex>\mathbb{E}\left[1\left\{K_j^{(b)}\leq\alpha\gamma\right\}\right] \leq \mathbb{P}\left[\left.P_j^{(b)}\leq\alpha\gamma\right|S\subseteq \tilde{S}^{(b)}\right].</tex>

Версия 11:45, 12 мая 2010

[Глоссарий статистических терминов ISI]


\mathbb{E}\left[1\left\{K_j^{(b)}\leq\alpha\gamma\right\}\right] \leq \mathbb{P}\left[\left.P_j^{(b)}\leq\alpha\gamma\right|S\subseteq \tilde{S}^{(b)}\right].

\mathbb{E}\left[1\left\{K_j^{(b)}\leq\alpha\gamma\right\}\right] = 1\cdot\mathbb{P}\left[K_j^{(b)}\leq\alpha\gamma\right] + 0\cdot\mathbb{P}\left[K_j^{(b)}>\alpha\gamma\right] = \mathbb{P}\left[K_j^{(b)}\leq\alpha\gamma\right] = \{\text{using the definition of } K_j^{(b)} \text{ and complete probability formula}\} =

 = \mathbb{P}\left[\left.P_j^{(b)}\leq\alpha\gamma\right|S\subseteq \tilde{S}^{(b)}\right] + \mathbb{P}\left[1\leq\alpha\gamma\left|S\not\subseteq \tilde{S}^{(b)}\right.\right] = \mathbb{P}\left[\left.P_j^{(b)}\leq\alpha\gamma\right|S\subseteq \tilde{S}^{(b)}\right]

\mathbb{P}\left[\left.P_j^{(b)}\leq\alpha\gamma\right|S\subseteq \tilde{S}^{(b)}\right] = \mathbb{P}\left[\left.\tilde{P}_j^{(b)}\leq\frac{\alpha\gamma}{\left|\tilde{S}^{(b)}\right|}\right|S\subseteq \tilde{S}^{(b)}\right]

\mathbb{P}\left[\left.P_j^{(b)}\leq\alpha\gamma\right|S\subseteq \tilde{S}^{(b)}\right] = \frac {\alpha\gamma} { \left| \tilde{S}^{(b)} \right| }.

\tilde{P}_j^{(b)}

S\subseteq \tilde{S}^{(b)}

\beta_j=0

\tilde{S}^{(b)}

\tilde{P}_j^{(b)}\leq\frac{\alpha\gamma}{\left|\tilde{S}^{(b)}

Личные инструменты