2

I have read a description of R package and find the following:

"Evaluate the density of the fitted model at (2.747, 0.1467, 0.13, 0.05334)".

I do not understand what the author mean by evaluate the density, and what are the nubmers to evalue the density at?

What is the benifit of evaluating the density function of a fitted model and how to specify the number which the density need to be evaluated at?

The code source is from the "vineclust" from github.

oezgesahin.github.io/vineclust/

Kindly note that this is not an R code questions. The link of the code

Alice
  • 640

1 Answers1

3

"Evaluate" here is used in the "to calculate the numerical value of" dictionary meaning. What the author of the code means is that they plug in those points to the distribution function to see what probability densities it returns.

What is the benefit of evaluating the density function

In many cases, the point of fitting the density function is so that you can later use it to make predictions in terms of the probability densities.

what are the numbers to evaluate the density at

Well, the numbers you are interested in learning the probability density for them. Usually, you know the points beforehand, because your purpose was to calculate those results for the points.

Tim
  • 138,066
  • Thanks a lot for your help. I think that evaluation the density of fitted model can be done for regression model only. However, this is a clustering method. And the evaluation function return the loglikelihood. Do you think the author aims to evaluate that these points belong to a specific cluster. – Alice Jun 27 '22 at 12:34
  • 1
    @Alice it could be used like this, but the code you linked to doesn't seem to be doing that because it evaluates the "whole" distribution rather than marginal distributions for the clusters. – Tim Jun 27 '22 at 13:07
  • Thanks. But how they compute the density at specific values? I mean that the pdf at specific value is zero. – Alice Jun 27 '22 at 17:24
  • I think they plug these points to the cdf not the pdf. Is that correct? – Alice Jun 27 '22 at 17:31
  • 1
    @Alice probability is zero, probability density is not. – Tim Jun 27 '22 at 18:23