Hi,
I'm currently trying to get into Mahout in order to get some Naive Bayes
classification running using Java. Now I'm not sure if I've overlooked
something, but I wasn't able to find simple examples, which were also
somewhat current, as I'm also trying to use the latest release (0.12).
The biggest problem seems to be that I should implement everything in
Java, as most examples and tutorials I've found so far are using the
Mahout command line interface at least for some steps, e.g. for doing
the training. I'm not sure why it seems so uncommon to implement
everything in Java. Is it a bad idea in general or is it just more
complicated?
Does maybe anybody know of examples or a book that could help me there?
Just in case it is easier for you to help on an actual problem:
Currently it seems like I'm stuck on writing the label indexes into the
file system using the method BayesUtils.writeLabelIndex(), as I'm not
sure on how to create an object of type Iterable<Pair<Text,
IntWritable>>, but I'm afraid this won't be the last problem I'll get
stuck on without a good example.
Thank you in advance and best regards,
Vawael
