<p>This also comes out of the discussion for <a href="https://github.com/mlpack/mlpack/pull/592" class="issue-link js-issue-link" data-url="https://github.com/mlpack/mlpack/issues/592" data-id="143358633" data-error-text="Failed to load issue title" data-permission-text="Issue title is private">#592</a>.  <a href="https://github.com/kno10" class="user-mention">@kno10</a>'s suggestion is that we simply pick points randomly from the dataset as initial centroids.</p>

<p>This requires a bit of refactoring because the current initialization policy operates by setting the assignments of each point, instead of the location of each centroid.  So there will need to be a little bit of thought on how exactly to handle this, but probably some template metaprogramming to handle both cases is in order here (so that a user can both write initialization policies that give initial point assignments and also initial centroids).</p>

<p>I will do this as I have time.</p>

<p style="font-size:small;-webkit-text-size-adjust:none;color:#666;">&mdash;<br />You are receiving this because you are subscribed to this thread.<br />Reply to this email directly or <a href="https://github.com/mlpack/mlpack/issues/596">view it on GitHub</a><img alt="" height="1" src="https://github.com/notifications/beacon/AJ4bFH-0MMyLRQyr_katPP6Vvu2iKttvks5pxK69gaJpZM4H5J1C.gif" width="1" /></p>
<div itemscope itemtype="http://schema.org/EmailMessage">
<div itemprop="action" itemscope itemtype="http://schema.org/ViewAction">
  <link itemprop="url" href="https://github.com/mlpack/mlpack/issues/596"></link>
  <meta itemprop="name" content="View Issue"></meta>
</div>
<meta itemprop="description" content="View this Issue on GitHub"></meta>
</div>