mahout-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Dunning <>
Subject Re: What about a universal input data handling mechanism for Mahout?
Date Thu, 28 Jul 2011 19:13:37 GMT
CSVRecordFactory is a class, not a program.  As such, it is not limited to
one situation or the other.

You can use it in several contexts, but you will have to somehow get it to
read the field headers in the map-reduce context.  This is slightly tricky
because splits other than the initial one will not contain the first line of
the file.

On Thu, Jul 28, 2011 at 7:57 AM, Xiaobo Gu <> wrote:

> I am not so familiar with the MapReduce programming style now, but I
> think CsvRecordFactory is designed to run locally, not for map-reduce,
> am I right?

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message