Hi All,
I am newbie using apache mahout and I don't know how to train and test
mahout naive bayes classifier with a set of text files. In my scenario I
have five directories labeled as text categories with its files into
them. They are organized as follow:

*Training dataSet:*

- Category1
    - file1.txt
    - file2.txt
    ...

- Category2
    - file10.txt
    - file11.txt
    ....

*Test dataset:*

 - Category1
    - file1.txt
    - file2.txt
    ...

- Category2
    - file10.txt
    - file11.txt
    ....

I would like to train and test Naive Bayes classifier using these two
datasets (train and test).
Questions:

1- What are the necessary steps to do that ?
2- How can I collect statics informations as weka shows ?
Cheers,
Felipe Ferreira.

Look at the Naive Bayes example in examples/bin/asf-examples.sh.

You would skip the first stage where it parses email files, and
instead just create Hadoop SequenceFiles with a key of the filename
and the value is a Text field with the contents of the value.

On Wed, Feb 15, 2012 at 7:38 AM, Felipe Ferreira
<[EMAIL PROTECTED]> wrote:
> Hi All,
>
>
> I am newbie using apache mahout and I don't know how to train and test
> mahout naive bayes classifier with a set of text files. In my scenario I
> have five directories labeled as text categories with its files into
> them. They are organized as follow:
>
> *Training dataSet:*
>
> - Category1
>    - file1.txt
>    - file2.txt
>    ...
>
> - Category2
>    - file10.txt
>    - file11.txt
>    ....
>
> *Test dataset:*
>
>  - Category1
>    - file1.txt
>    - file2.txt
>    ...
>
> - Category2
>    - file10.txt
>    - file11.txt
>    ....
>
> I would like to train and test Naive Bayes classifier using these two
> datasets (train and test).
>
>
> Questions:
>
> 1- What are the necessary steps to do that ?
> 2- How can I collect statics informations as weka shows ?
>
>
> Cheers,
>
>
> Felipe Ferreira.

--
Lance Norskog
[EMAIL PROTECTED]

I take that back. Follow the Naive Bayes example in
examples/bin/asf-examples.sh. You'll have to examine the files in each
stage to figure out exactly what to pass in.

On Wed, Feb 15, 2012 at 7:50 PM, Lance Norskog <[EMAIL PROTECTED]> wrote:
> Look at the Naive Bayes example in examples/bin/asf-examples.sh.
>
> You would skip the first stage where it parses email files, and
> instead just create Hadoop SequenceFiles with a key of the filename
> and the value is a Text field with the contents of the value.
>
> On Wed, Feb 15, 2012 at 7:38 AM, Felipe Ferreira
> <[EMAIL PROTECTED]> wrote:
>> Hi All,
>>
>>
>> I am newbie using apache mahout and I don't know how to train and test
>> mahout naive bayes classifier with a set of text files. In my scenario I
>> have five directories labeled as text categories with its files into
>> them. They are organized as follow:
>>
>> *Training dataSet:*
>>
>> - Category1
>>    - file1.txt
>>    - file2.txt
>>    ...
>>
>> - Category2
>>    - file10.txt
>>    - file11.txt
>>    ....
>>
>> *Test dataset:*
>>
>>  - Category1
>>    - file1.txt
>>    - file2.txt
>>    ...
>>
>> - Category2
>>    - file10.txt
>>    - file11.txt
>>    ....
>>
>> I would like to train and test Naive Bayes classifier using these two
>> datasets (train and test).
>>
>>
>> Questions:
>>
>> 1- What are the necessary steps to do that ?
>> 2- How can I collect statics informations as weka shows ?
>>
>>
>> Cheers,
>>
>>
>> Felipe Ferreira.
>
>
>
> --
> Lance Norskog
> [EMAIL PROTECTED]

--
Lance Norskog
[EMAIL PROTECTED]

Thank you Lance Norskog. I am going to try understand this example and to
do the same steps in my project. After that, I would like to compare mahout
tests results with weka tests results because I already performed a
training session using the same datasets in weka and I got 89 % of
correctly classified instances. Do you think I would get the same
percentage in mahout ?
On Wed, Feb 15, 2012 at 11:55 PM, Lance Norskog <[EMAIL PROTECTED]> wrote:

> I take that back. Follow the Naive Bayes example in
> examples/bin/asf-examples.sh. You'll have to examine the files in each
> stage to figure out exactly what to pass in.
>
> On Wed, Feb 15, 2012 at 7:50 PM, Lance Norskog <[EMAIL PROTECTED]> wrote:
> > Look at the Naive Bayes example in examples/bin/asf-examples.sh.
> >
> > You would skip the first stage where it parses email files, and
> > instead just create Hadoop SequenceFiles with a key of the filename
> > and the value is a Text field with the contents of the value.
> >
> > On Wed, Feb 15, 2012 at 7:38 AM, Felipe Ferreira
> > <[EMAIL PROTECTED]> wrote:
> >> Hi All,
> >>
> >>
> >> I am newbie using apache mahout and I don't know how to train and test
> >> mahout naive bayes classifier with a set of text files. In my scenario I
> >> have five directories labeled as text categories with its files into
> >> them. They are organized as follow:
> >>
> >> *Training dataSet:*
> >>
> >> - Category1
> >>    - file1.txt
> >>    - file2.txt
> >>    ...
> >>
> >> - Category2
> >>    - file10.txt
> >>    - file11.txt
> >>    ....
> >>
> >> *Test dataset:*
> >>
> >>  - Category1
> >>    - file1.txt
> >>    - file2.txt
> >>    ...
> >>
> >> - Category2
> >>    - file10.txt
> >>    - file11.txt
> >>    ....
> >>
> >> I would like to train and test Naive Bayes classifier using these two
> >> datasets (train and test).
> >>
> >>
> >> Questions:
> >>
> >> 1- What are the necessary steps to do that ?
> >> 2- How can I collect statics informations as weka shows ?
> >>
> >>
> >> Cheers,
> >>
> >>
> >> Felipe Ferreira.
> >
> >
> >
> > --
> > Lance Norskog
> > [EMAIL PROTECTED]
>
>
>
> --
> Lance Norskog
> [EMAIL PROTECTED]
>

I have not used the Weka tools for this.

On Thu, Feb 16, 2012 at 9:44 AM, Felipe Ferreira
<[EMAIL PROTECTED]> wrote:
> Thank you Lance Norskog. I am going to try understand this example and to
> do the same steps in my project. After that, I would like to compare mahout
> tests results with weka tests results because I already performed a
> training session using the same datasets in weka and I got 89 % of
> correctly classified instances. Do you think I would get the same
> percentage in mahout ?
> On Wed, Feb 15, 2012 at 11:55 PM, Lance Norskog <[EMAIL PROTECTED]> wrote:
>
>> I take that back. Follow the Naive Bayes example in
>> examples/bin/asf-examples.sh. You'll have to examine the files in each
>> stage to figure out exactly what to pass in.
>>
>> On Wed, Feb 15, 2012 at 7:50 PM, Lance Norskog <[EMAIL PROTECTED]> wrote:
>> > Look at the Naive Bayes example in examples/bin/asf-examples.sh.
>> >
>> > You would skip the first stage where it parses email files, and
>> > instead just create Hadoop SequenceFiles with a key of the filename
>> > and the value is a Text field with the contents of the value.
>> >
>> > On Wed, Feb 15, 2012 at 7:38 AM, Felipe Ferreira
>> > <[EMAIL PROTECTED]> wrote:
>> >> Hi All,
>> >>
>> >>
>> >> I am newbie using apache mahout and I don't know how to train and test
>> >> mahout naive bayes classifier with a set of text files. In my scenario I
>> >> have five directories labeled as text categories with its files into
>> >> them. They are organized as follow:
>> >>
>> >> *Training dataSet:*
>> >>
>> >> - Category1
>> >>    - file1.txt
>> >>    - file2.txt
>> >>    ...
>> >>
>> >> - Category2
>> >>    - file10.txt
>> >>    - file11.txt
>> >>    ....
>> >>
>> >> *Test dataset:*
>> >>
>> >>  - Category1
>> >>    - file1.txt
>> >>    - file2.txt
>> >>    ...
>> >>
>> >> - Category2
>> >>    - file10.txt
>> >>    - file11.txt
>> >>    ....
>> >>
>> >> I would like to train and test Naive Bayes classifier using these two
>> >> datasets (train and test).
>> >>
>> >>
>> >> Questions:
>> >>
>> >> 1- What are the necessary steps to do that ?
>> >> 2- How can I collect statics informations as weka shows ?
>> >>
>> >>
>> >> Cheers,
>> >>
>> >>
>> >> Felipe Ferreira.
>> >
>> >
>> >
>> > --
>> > Lance Norskog
>> > [EMAIL PROTECTED]
>>
>>
>>
>> --
>> Lance Norskog
>> [EMAIL PROTECTED]
>>

--
Lance Norskog
[EMAIL PROTECTED]