I'm reading The Elements of Statistical Learning. I have a question about the curse of dimensionalit



Answered question


I'm reading The Elements of Statistical Learning. I have a question about the curse of dimensionality.
In section 2.5, p.22:
Consider N data points uniformly distributed in a p-dimensional unit ball centered at the origin. suppose we consider a nearest-neighbor estimate at the origin. The median distance from the origin to the closest data point is given by the expression:
d ( p , N ) = ( 1 1 2 1 / N ) 1 / p .
For N=500, p=10, d ( p , N ) 0.52, more than halfway to the boundary. Hence most data points are closer to the boundary of the sample space than to any other data point.
I accept the equation. My question is, how we deduce this conclusion?

Answer & Explanation



Beginner2022-06-22Added 14 answers

This is the exercise 2.3 that they refer to.

Probability Distribution Function is mentioned in the PDF.

Cumulative Distribution Function, or CDF.

The former is the derivative of the latter since continuous distributions are what we are thinking about.
The volume of a ball of radius r in R p is ω p r p ,, where ωp is a constant depending only on p, the value indicated by shorthand
ω p = π p / 2 ( p / 2 ) ! ..
As a result, the probability that a point, taken uniformly in the unit ball, is within distance x of the origin is the volume of that ball divided by the volume of the unit ball. The factors of ω p cancel, so we get CDF
F ( x ) = x p ,    0  x  1..
The corresponding PDF is the derivative,
f ( x ) = p x p  1 ,    0  x  1..
From page 150, section 4.6 of Introduction to Mathematical Statistics by Hogg and Craig, we are told that the marginal (individual) PDF for y 1 ,, the smallest order statistic (the minimum) of n points with CDF F and PDF f is
g ( y ) = n ( 1  F ( y ) ) n  1 f ( y ) ..
In our case that gives
g ( y ) = n ( 1  y p ) n  1 p y p  1 ,,
It is easily incorporated to provide the CDF
G ( y ) = 1  ( 1  y p ) n ..
The expected value of y, or the mean, is a confusing integral. Instead, in the case of a continuous variable, the median is simply defined as the value of the random variable y such that G(y)=1/2. The probability of having a minimum less than the median is 50% if you repeated the experiment, and the probability of receiving a minimum greater than the median is also 50%. The median and mean are probably quite close for the conventional bell curve. I'm not sure if the median and mean are necessary close to one another in this case because the polynomial in question is constrained to a limited interval.I don't understand how you could ever read this book without having taken a complete semester of calculus-based quantitative statistics.
Solve G(y)=1/2, you get their expression.

Do you have a similar question?

Recalculate according to your conditions!

Ask your question.
Get an expert answer.

Let our experts help you. Answer in as fast as 15 minutes.

Didn't find what you were looking for?