What is the Local Outlier Factor

Continuing in our series on anomaly detection, let's build off the last video on k-nearest neighbors and talk about another common technique, the local outlier factor. All in under 5 minutes of course!

[@aryashahdi2790]

salute to this dude for the clarity of his explanations

[@Anjali.Jivani]

Amazing way of explaining

[@kaynkayn9870]

Somehow you make these videos extremely informative in only 5 minutes. What a legend.

[@mahmoudel-bahnasawi2809]

This series is truly unique; please keep it going.

[@baruite]

Tellement bien expliquée! merci

[@AynazAbdollahzadeh]

I was super lost thanks for explaining it amazingly!

[@CP-tq1ue]

Thanks!

[@space-time-somdeep]

Please continue the series sir

[@tehreemqasim2204]

Very well explained thank you

[@riccardorossi5224]

Hi, I wanted to ask you a question. I understood your reasoning by comparing circles and indicating as an outlier if the point of my observation is larger than that of its neighbors. But in reality it is wrong to say that it is an outlier because it has a higher density than the density of its neighbors. High density means he has samples very close to him, low density means he has samples very far from him. Therefore, the sample that is very far from the other samples, and therefore has a lower density, is an outlier. Tell me if you understand what I mean, if you can correct me you'll do me a favor.

[@AricLaBarr]

No problem at all! In reality the circles represent the opposite of density and more reachability. So the larger the circles mean the larger the reachability (inverse of density). That is what makes the large circles more likely to be outliers! Hope this helps!

[@riccardorossi5224]

@@AricLaBarr Okey, Thanks again.

[@chrismawata8755]

At 1:50 the density is defined as the inverse of the average reachability ... Somehow the 'inverse' was ignored after that which flips the meaning of density after that point.

[@philtoa334]

Nice.