[
https://issues.apache.org/jira/browse/MADLIB-1159?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16168251#comment-16168251
]
Brian Dolan commented on MADLIB-1159:
-------------------------------------
Sure!
Map Documents, using 1 indexing (but that's how counting works :) )
doc_name | id
-------------------------------------------------|---
do androids dream of electric sheep | 1
da vinci code book review | 2
Map features
ftr | id
----------------|--
rachel | 1
andy | 2
hands | 3
vapid | 4
uninspired | 5
nauseating | 6
inane | 7
Then the matrix would be 2 x 7. This one looks boring, but you get the picture.
1 1 1 0 0 0 0
0 0 0 1 1 1 1
Does that help?
> Provide examples for common sparse matrix cases
> -----------------------------------------------
>
> Key: MADLIB-1159
> URL: https://issues.apache.org/jira/browse/MADLIB-1159
> Project: Apache MADlib
> Issue Type: Documentation
> Reporter: Brian Dolan
>
> A fairly common table structure is of the form `key1, key2, value` like a
> triples in a graph. These are often not normalized.
> It would be useful to provide an example of transforming this class of tables
> into a sparse matrix. Perhaps an example dataset could be a term-document
> matrix.
> TABLE doc_term;
> document, term, freq
> "do androids dream of electric sheep", "rachel", 75
> "do androids dream of electric sheep", "andy", 56
> "do androids dream of electric sheep", "hands", 128
> "da vinci code book review", "vapid",1326
> "da vinci code book review", "uninspired",265
> "da vinci code book review", "nauseating",879293
> "da vinci code book review", "inane",471
> Into a sparse matrix table of documents by features.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
