The Lounge is rated Safe For Work. If you're about to post something inappropriate for a shared office environment, then don't post it. No ads, no abuse, and no programming questions. Trolling, (political, climate, religious or whatever) will result in your account being removed.
After a quick view I can see nothing wrong with the article.
I think it's the subject, statistics just isn't for everyone.
It's also fairly advanced, which sorts away quite a few more readers.
Compare with CodeWitch's articles, she's mass producing advanced articles that needs commitment to read and understand, but gets the article of the month for a fairly simple straightforward article on adding list functionality for collections.
If it's points you're after you need to write articles for the broad mass of readers.
<edit> That said, I have put your article on my reading list</edit>
It looks like a great article, but it's just not a subject that I need to read about.
Can't find anything wrong with it either.
Anything that is unrelated to elephants is irrelephant Anonymous - The problem with quotes on the internet is that you can never tell if they're genuine Winston Churchill, 1944 - Never argue with a fool. Onlookers may not be able to tell the difference. Mark Twain
I've just looked at the stats for your article. It shows that you had a peak of a little over 800 people looking at it in one day. Since then, there has been a steady decline. So why was there a peak? There would have been a peak because the article was visible on the home page at this point so it was something that was "in your face" for people coming in to the site. Given site traffic, why has this not been higher? If you think of the home page as being like a shop front, the title and description you give your article acts as the packaging. If the packaging isn't eye catching, people aren't going to bother with it. Your title needs to scream "look at me, I'm something interesting" and your description needs to tell people, "this is what I'm about and this is why YOU NEED to know about me". If you don't catch the readers attention instantly, you are competing with all the other articles on the home page. Once the article falls off the home page, your title, description, categorisation and tags become vital.
In my opinion, when you use what an article is literally about in the title you'll only attract the type of people interested in that subject. I see "optimal k-means clustering" and I think "that sounds interesting but complex. Don't have time, might bookmark for a rainy day." So for traffic, it is probably a better idea to use the title to explain what this article practically accomplishes and maybe include the details in the sub-header description.
So instead of "A Gentle Introduction to Optimal K-Means Clustering" something like "A Gentle Introduction to Multi-Objective AIs" with the sub-header "An optimal k-means clustering algorithm". That grabs my attention more. A multi-purpose AI? Hmmm, that might be useful in -insert project or situation here-.
Such targeted articles still won't generate the traffic of more general-purpose stuff but I know from my own habits that if I know why an article is useful I'm much more likely to read it even if I don't fully understand the topics because I now have a motivation to invest at least some time in doing the background research to understand those topics at a basic level
The primary subject is still "Optimal K-Means Clustering Algorithm" which before I clicked and read a bit of your article because of this post, I had zero idea what on Earth that even was. If I was busy and skimming for articles to read later from the front page I would probably pass to be honest. I think if you somehow hint in the title it relates to AI you'd get way more interest.
Edit: Of course I don't really do AI stuff, so maybe that's a common algorithm? This is all just my perspective as an AI noob
I didn't notice it mentioned but what's the Big-Oh performance of the sub-optimal k-means clustering algorithm? O(k)?
Quote Selected Text
The complexity of the sub-optimal (!) k-means algorithm is typically very high (e.g. it's NP-hard).
That's actually why I've used the number of algorithm optimizations such as k-means++ initialization algorithm, thoroughly discussed in this article, as well as the ability of the initialization process to produce the number of initial clusters prior to performing the actual clustering, and that really helps to reduce the computational complexity of the k-means algorithm.