Very good. Thanks again! I’m sure future visitors to this page will be grateful for the clarification.

Thank you for your feedback. I updated my previous comment. It is the latest developmental version of factoextra that returns H instead of 1-H (https://github.com/kassambara/factoextra/blob/master/NEWS.md).

You can install it as follow:

if(!require(devtools)) install.packages("devtools")
devtools::install_github("kassambara/factoextra")

I’m submitting it to CRAN now.

Thanks for producing so much useful content! Your website is great.

However, you said in a comment here that the get_clust_tendency() function in version 1.0.5 of ‘factoextra’ generates H, rather than 1 – H. This is incorrect, as far as I can tell, because I am using ‘factoextra’ version 1.0.5, and get output values from get_clust_tendency() which are virtually identical to the output of the hopkins() function from ‘clustertend’ (which, as you know, outputs 1 – H). I have tested this with different datasets, and it seems the get_clust_tendency() function in version 1.0.5 of ‘factoextra’ generates 1 – H, not H.

Please consider providing a proper explanation in the main text of this tutorial, of when and how the get_clust_tendency() function from ‘factoextra’ was changed.

Many people rely on the content of your website, and not providing a proper explanation has resulted in quite a lot of confusion, as you can see from the comments here and discussions elsewhere too: https://stats.stackexchange.com/questions/332651/validating-cluster-tendency-using-hopkins-statistic

Many thanks again for the useful content!

If v.1.0.5 and beyond computes H, then ;

res = get_clust_tendency(
data = data,
n = nrow(data)-1,
graph = F
)

res$hopkins_stat = 0.1815219

——

Here data is the standardised Iris dataset. How do we proceed?

https://cran.r-project.org/web/packages/factoextra/factoextra.pdf

it is written below 0.5 is highly clusterable
but in this it is 0.8
how it is highly clusterable

Please make sure you have the latest developmental version of factoextra, which computes H.

You can install it as follow:

if(!require(devtools)) install.packages("devtools")
devtools::install_github("kassambara/factoextra")

I am also questioning if the get_clust_tendency from devtools::install_github(“kassambara/factoextra”) produces 1-H instead of H.

There is also a discussion here: https://stats.stackexchange.com/questions/332651/validating-cluster-tendency-using-hopkins-statistic

Can the author please confirm?

Hi,

R is asking if you want to update old packages. Type 20 in R console to ignore these updates and try.