Let $X$ denote a random variable. Then from a rigorous mathematical perspective (books such as Durrett, Feller, Kolmogorov, etc.),

$X$ is a function. $X: \Omega \to \mathbb{R}^n$.

Domain of the function is the sample space, $\Omega$

Range is a value in $\mathbb{R}^n$.

In supervised learning, let $X, Y$ denote the random variable corresponding to the data $x_n \in \mathbb{R}^n$ and target $y_n \in \mathbb{R}$ respectively.

Then $X$ maps from a sample space into a piece of data $x_n$. And $Y$ maps from a sample space into a piece of target, $y_n$.

With this, the notation, such as $p_{X|Y}(x_n|y_n) = \Pr[X = x_n| Y = y_n]$

is completely well defined.

Then we can start talking about things like logistic regression, i.e.,

$p_{X|Y}(x_n|y_n) = \Pr[X = x_n| Y = y_n] = \text{logit}(y_nw^Tx_n) \in (0,1)$


But what is the sample space $\Omega$? Any example would be much appreciated!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.