Problem with GSP

[Quanzhi LIU]

Hi List,

I need an Arff file to use with GSP algorithm.since there's no exemple of date in weka that can be used,the existing Arff files such as weather.nominal.arff does not really work, i don't know what format of data should be used. what i open a file, the methode of "generalized sequential pattern" always can't be chosen.

I have searched in the internet at least a week, but still can't find the answer.

 Can anyone help?
Thanks.
Quanzhi
--
Informatique et Gestion
Polytech, Montpellier 2
Montpellier, France
                                                                 


_______________________________________________
Wekalist mailing list
Send posts to: Wekalist@...
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

[Peter Reutemann-3]

> I need an Arff file to use with GSP algorithm.since there's no exemple of
> date in weka that can be used,the existing Arff files such as
> weather.nominal.arff does not really work, i don't know what format of data
> should be used. what i open a file, the methode of "generalized sequential
> pattern" always can't be chosen.
>
> I have searched in the internet at least a week, but still can't find the
> answer.

Searching the Wekalist archives for "GeneralizedSequencePattern" would
have been sufficient:
  https://list.scms.waikato.ac.nz/mailman/htdig/wekalist/2007-October/011433.html

NB: GSP was a contribution, so it's probably a good idea to contact the author.

Cheers, Peter
--
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/           Ph. [phone]

_______________________________________________
Wekalist mailing list
Send posts to: Wekalist@...
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

[Quanzhi LIU]

Thanks a lot, Peter!

2009/6/29 Peter Reutemann <fracpete@...>

> I need an Arff file to use with GSP algorithm.since there's no exemple of
> date in weka that can be used,the existing Arff files such as
> weather.nominal.arff does not really work, i don't know what format of data
> should be used. what i open a file, the methode of "generalized sequential
> pattern" always can't be chosen.
>
> I have searched in the internet at least a week, but still can't find the
> answer.

Searching the Wekalist archives for "GeneralizedSequencePattern" would
have been sufficient:
 https://list.scms.waikato.ac.nz/mailman/htdig/wekalist/2007-October/011433.html

NB: GSP was a contribution, so it's probably a good idea to contact the author.

Cheers, Peter
--
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/           Ph. [phone]

_______________________________________________
Wekalist mailing list
Send posts to: Wekalist@...
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

--
Informatique et Gestion
                                                             Polytech, Montpellier 2
                                                                   Montpellier, France
                                                                 


_______________________________________________
Wekalist mailing list
Send posts to: Wekalist@...
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

[Sebastian Beer]

Hi there,

On Mon, Jun 29, 2009 at 11:32 AM, Quanzhi LIU<vickyquanzhi@...> wrote:
>  Can anyone help?

I've attached a little explanation of the GSP algorithm (which now is
archived with regard to future questions).

Hope this helps a bit.

Best regards,
Sebastian



--------------------

Consider the following minimalistic data set:

-----
@relation sequential_test_set

@attribute day {1, 2, 3}
@attribute power consumption {gas=base, gas=peak}
@attribute wind speed {wind=calm, wind=breeze}

@data
1,power=base,wind=calm
1,power=peak,wind=breeze
2,power=base,wind=calm
2,power=peak,wind=breeze
3,power=base,wind=calm
3,power=peak,wind=breeze
----

There are 3 subsets describing the power consumption and wind speed
over the day; e.g. the first line could be interpreted as "on day one
the power cunsumption corresponded to a base load level and the wind
speed was calm". The algorithm extracts these three subsets, deletes
the attribute identifying the subsets (in this case "day"), and
extracts all existing sequences meeting a previously set minimum
support threshold. The output looks like the following:

---------------------------------------------------------------------------------------------------------

- Number of cycles performed: 4
- Total number of frequent sequences: 15

- Algorithm started at: 2007-04-23, 18:25:48
- Algorithm ended at: 2007-04-23, 19:27:50

- Frequent Sequences Details (filtered)
-----------------------------------------------------------------

- 1-sequences ------------------------------------------

[1] <{power=base}> (3)
[2] <{power=peak}> (3)
[3] <{wind=calm}> (3)
[4] <{wind=breeze}> (3)

- 2-sequences ------------------------------------------

[1] <{power=base}{power=peak}> (3)
[2] <{power=base}{wind=breeze}> (3)
[3] <{power=base,wind=calm}> (3)
[4] <{wind=calm}{power=peak}> (3)
[5] <{wind=calm}{wind=breeze}> (3)
[6] <{power=peak,wind=breeze}> (3)

- 3-sequences ------------------------------------------

[1] <{power=base}{power=peak,wind=breeze}> (3)
[2] <{power=base,wind=calm}{power=peak}> (3)
[3] <{power=base,wind=calm}{wind=breeze}> (3)
[4] <{wind=calm}{power=peak,wind=breeze}> (3)

- 4-sequences ------------------------------------------

[1] <{power=base,wind=calm}{power=peak,wind=breeze}> (3)

---------------------------------------------------------------------------------------------------------

Generally, an exemplary sequence looks like

<{event1,event2}{event3}{event4,event5}>

where the curly braces define an element of the sequence.

At the moment, there are 3 options available:

---
dataSeqID -- The attribute number representing the data sequence ID
(in the example "day").

filterAttributes -- The attribute numbers (e.g. "0, 1") used for
result filtering. Only sequences containing the specified attributes
in each of their elements/itemsets will be output. -1 prints all.

minSupport -- Minimum support threshold.
---

_______________________________________________
Wekalist mailing list
Send posts to: Wekalist@...
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html