Skip to main content
13 votes

Why Gaussian Process Regression (GPR) is non-parametric?

Generally speaking, with the way people use the term nonparametric to decribe a model, if you had more data points, there would be more parameters, generally without an upper limit. It doesn't mean &...
Glen_b's user avatar
  • 290k
9 votes
Accepted

Why Gaussian Process Regression (GPR) is non-parametric?

The points you raised apply to K Nearest Neighbors Regression too. Let me start by rephrasing what you wrote in terms of K-nearest neighbors regression: """ Given that K nearest ...
John Madden's user avatar
  • 5,660
4 votes
Accepted

Maximum value of Friedman's randomized block statistic

The general form of the Friedman statistic using the ranks $R_1,...,R_k$ is: $$Q(\mathbf{R}) = \frac{12n}{k(k+1)} \sum_{i=1}^k \bigg( R_i - \frac{k+1}{2} \bigg)^2.$$ The statistic is maximised when ...
Ben's user avatar
  • 133k
3 votes
Accepted

Cross-validated bandwidth for the derivative of the function with local quadratic estimation

Section 5.3 of Boasiako and Staudenmayer 2022 is relevant to our discussion. It confirms your assumption that "that the optimal bandwidth for the derivative is not necessarily the same as the ...
John Madden's user avatar
  • 5,660
2 votes

Cross-validated bandwidth for the derivative of the function with local quadratic estimation

Example 1 You can not use a general rule of thumb that converts the bandwith for estimating $g(x)$ to a bandwidth for estimating $g'(x)$. The reason is that it depends on the situation and sometimes ...
Sextus Empiricus's user avatar
1 vote

Gaining additional information after Kruskal-Wallis by further comparing groups using Wilcoxon? (when two-way ANOVA is not possible)

I will start by assuming that when you say "Wilcoxon", you mean Wilcoxon Rank Sum test (WRSt), aka Mann-Whitney U test (MWUt). I will also take into account the comment you made about ...
jginestet's user avatar
  • 5,294
1 vote

What is the weak side of decision trees?

Some weakness (which is surprisingly rarely mentioned) of tree-based algorithms for regression is: They are by design unable to predict values outside the seen value range. So they can not extrapolate....
tturbo's user avatar
  • 111

Only top scored, non community-wiki answers of a minimum length are eligible