Yes, I just mean I think we have heard about this casting issue for a while and not sure it was fixed up in 0.5. I imagine it is only a shallow problem, but I also wasn't sure if anyone was looking after k-means since I am not sure I recall anyone looking into this.
Let me file a JIRA to track this. (Apropos of the talk of needing a roadmap ... these are the kinds of smallish issues that make up a fine de facto roadmap for the near future! plenty to do even right now.) On Thu, Jun 9, 2011 at 4:45 PM, Hector Yee <[email protected]> wrote: > I was following the book examples and k means , dirichlet and lda all have > this casting problem. It may be a Mac issue not sure . I suspect it may be > seq2sparse messing up the inputs, maybe wrong version. It outputs the > regular part-r-* but the lda driver expects a file called data. > > Sent from my iPad > > On Jun 9, 2011, at 7:40 AM, Mark <[email protected]> wrote: > > > Forgot to mention... great book :) > > > > On 6/9/11 7:30 AM, Mark wrote: > >> KMeans is busted? What do you mean by this? The algorithm simply won't > work or just the reuters example? > >> > >> Thanks > >
