I searched the internet to find out about the effects of outliers on k-means clustering, but could not find any useful source. Does it cause bad clusters to be formed? If so, how?

(Please also provide a simple example with data such as a list of integers so that my doubt can be clarified.)

  • 2
    $\begingroup$ The answer likely hinges on what you really mean by "outlier" and "bad cluster." Ordinarily an "outlier" is just that--a data value that according to some measure is "far" from typical data values. In itself it's neither "good" nor "bad"--it just is. One might suppose that a sufficiently extreme outlier or group of outliers ought to be in its own cluster, so are you asking whether common K-means algorithms will tend to produce that kind of solution? $\endgroup$ – whuber Jun 28 '17 at 17:37
  • 1
    $\begingroup$ I think the question is too broad, please narrow it. $\endgroup$ – ttnphns Jun 28 '17 at 20:19