You can subscribe to this list here.
| 2001 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
(7) |
Aug
(3) |
Sep
(8) |
Oct
(9) |
Nov
(11) |
Dec
(54) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2002 |
Jan
(19) |
Feb
(41) |
Mar
(48) |
Apr
(42) |
May
(77) |
Jun
(43) |
Jul
(58) |
Aug
(55) |
Sep
(83) |
Oct
(113) |
Nov
(137) |
Dec
(40) |
| 2003 |
Jan
(95) |
Feb
(119) |
Mar
(124) |
Apr
(122) |
May
(99) |
Jun
(89) |
Jul
(101) |
Aug
(87) |
Sep
(80) |
Oct
(97) |
Nov
(142) |
Dec
(71) |
| 2004 |
Jan
(96) |
Feb
(105) |
Mar
(70) |
Apr
(87) |
May
(98) |
Jun
(146) |
Jul
(100) |
Aug
(70) |
Sep
(54) |
Oct
(81) |
Nov
(106) |
Dec
(108) |
| 2005 |
Jan
(113) |
Feb
(117) |
Mar
(181) |
Apr
(285) |
May
(224) |
Jun
(266) |
Jul
(154) |
Aug
(290) |
Sep
(280) |
Oct
(258) |
Nov
(206) |
Dec
(211) |
| 2006 |
Jan
(199) |
Feb
(239) |
Mar
(331) |
Apr
(290) |
May
(316) |
Jun
(290) |
Jul
(209) |
Aug
(187) |
Sep
(198) |
Oct
(197) |
Nov
(262) |
Dec
(155) |
| 2007 |
Jan
(235) |
Feb
(233) |
Mar
(303) |
Apr
(409) |
May
(364) |
Jun
(276) |
Jul
(312) |
Aug
(185) |
Sep
(248) |
Oct
(288) |
Nov
(216) |
Dec
(196) |
| 2008 |
Jan
(263) |
Feb
(263) |
Mar
(339) |
Apr
(354) |
May
(218) |
Jun
(274) |
Jul
(328) |
Aug
(250) |
Sep
(216) |
Oct
(333) |
Nov
(344) |
Dec
(224) |
| 2009 |
Jan
(337) |
Feb
(362) |
Mar
(525) |
Apr
(500) |
May
(527) |
Jun
(454) |
Jul
(319) |
Aug
(453) |
Sep
(437) |
Oct
(367) |
Nov
(364) |
Dec
(332) |
| 2010 |
Jan
(397) |
Feb
(269) |
Mar
(305) |
Apr
(332) |
May
(410) |
Jun
(345) |
Jul
(271) |
Aug
(350) |
Sep
(291) |
Oct
(338) |
Nov
(229) |
Dec
(234) |
| 2011 |
Jan
(300) |
Feb
(366) |
Mar
(460) |
Apr
(273) |
May
(322) |
Jun
(247) |
Jul
(277) |
Aug
(326) |
Sep
(248) |
Oct
(247) |
Nov
(284) |
Dec
(285) |
| 2012 |
Jan
(224) |
Feb
(233) |
Mar
(333) |
Apr
(221) |
May
(228) |
Jun
(191) |
Jul
(148) |
Aug
(180) |
Sep
(119) |
Oct
(177) |
Nov
(174) |
Dec
(200) |
| 2013 |
Jan
(218) |
Feb
(299) |
Mar
(307) |
Apr
(356) |
May
(382) |
Jun
(301) |
Jul
(262) |
Aug
(157) |
Sep
(115) |
Oct
(140) |
Nov
(209) |
Dec
(131) |
| 2014 |
Jan
(115) |
Feb
(268) |
Mar
(344) |
Apr
(387) |
May
(303) |
Jun
(211) |
Jul
(247) |
Aug
(240) |
Sep
(244) |
Oct
(254) |
Nov
(315) |
Dec
(318) |
| 2015 |
Jan
(310) |
Feb
(220) |
Mar
(306) |
Apr
(233) |
May
(309) |
Jun
(234) |
Jul
(252) |
Aug
(127) |
Sep
(195) |
Oct
(184) |
Nov
(187) |
Dec
(261) |
| 2016 |
Jan
(224) |
Feb
(218) |
Mar
(194) |
Apr
(316) |
May
(273) |
Jun
(223) |
Jul
(76) |
Aug
(134) |
Sep
(200) |
Oct
(287) |
Nov
(290) |
Dec
(240) |
| 2017 |
Jan
(201) |
Feb
(296) |
Mar
(270) |
Apr
(445) |
May
(412) |
Jun
(212) |
Jul
(270) |
Aug
(171) |
Sep
(203) |
Oct
(169) |
Nov
(167) |
Dec
(235) |
| 2018 |
Jan
(225) |
Feb
(179) |
Mar
(262) |
Apr
(195) |
May
(171) |
Jun
(136) |
Jul
(156) |
Aug
(165) |
Sep
(199) |
Oct
(140) |
Nov
(195) |
Dec
(84) |
| 2019 |
Jan
(87) |
Feb
(120) |
Mar
(143) |
Apr
(204) |
May
(145) |
Jun
(94) |
Jul
(154) |
Aug
(104) |
Sep
(173) |
Oct
(109) |
Nov
(171) |
Dec
(102) |
| 2020 |
Jan
(104) |
Feb
(116) |
Mar
(94) |
Apr
(172) |
May
(108) |
Jun
(91) |
Jul
(40) |
Aug
(53) |
Sep
(27) |
Oct
(44) |
Nov
(41) |
Dec
(53) |
| 2021 |
Jan
(82) |
Feb
(90) |
Mar
(62) |
Apr
(78) |
May
(32) |
Jun
(97) |
Jul
(34) |
Aug
(17) |
Sep
(53) |
Oct
(59) |
Nov
(76) |
Dec
(8) |
| 2022 |
Jan
(28) |
Feb
(20) |
Mar
(31) |
Apr
(24) |
May
(8) |
Jun
(28) |
Jul
(32) |
Aug
(12) |
Sep
(4) |
Oct
(24) |
Nov
(10) |
Dec
(5) |
| 2023 |
Jan
(27) |
Feb
(15) |
Mar
(6) |
Apr
(4) |
May
(8) |
Jun
(6) |
Jul
(13) |
Aug
(39) |
Sep
(4) |
Oct
(6) |
Nov
(8) |
Dec
(15) |
| 2024 |
Jan
(20) |
Feb
(16) |
Mar
(17) |
Apr
(22) |
May
(11) |
Jun
(14) |
Jul
(24) |
Aug
(4) |
Sep
(9) |
Oct
(2) |
Nov
(5) |
Dec
(9) |
| 2025 |
Jan
|
Feb
|
Mar
(12) |
Apr
(2) |
May
|
Jun
(2) |
Jul
|
Aug
|
Sep
(7) |
Oct
|
Nov
(7) |
Dec
|
| 2026 |
Jan
|
Feb
|
Mar
|
Apr
(1) |
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
(2) |
Nov
|
Dec
|
| S | M | T | W | T | F | S |
|---|---|---|---|---|---|---|
|
|
|
|
1
(25) |
2
(8) |
3
(45) |
4
(6) |
|
5
(15) |
6
(6) |
7
(16) |
8
(30) |
9
(5) |
10
(8) |
11
(8) |
|
12
(10) |
13
(17) |
14
(9) |
15
(19) |
16
(13) |
17
(27) |
18
(9) |
|
19
|
20
(35) |
21
(30) |
22
(13) |
23
(28) |
24
(16) |
25
(13) |
|
26
(5) |
27
(22) |
28
(22) |
29
(24) |
30
(16) |
|
|
|
From: Thomas D. <tw...@ne...> - 2009-04-30 21:45:39
|
2009/4/30 ashvin <ashvin.dhar(a)gmail.com> > Dear Weka users, > I have a dataset with nominal and numeric attributes. I want to convert the > nominal attributes to numeric. I used Nominaltobinay filter for converting > my nominal attribute which has 15 distinct values and this resulted in 15 > new columns in the dataset. > > Is there a better way of representing my data or converting the nominal to > numeric type? I want to represent the nominal attribute with a single > numeric value. > > > thanks, > ashvin. > > > _______________________________________________ > Wekalist mailing list > Send posts to: Wekalist(a)list.scms.waikato.ac.nz > 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html> > > hello, may I ask why you want to convert nominal attributes to numeric ones? As I understand, you want to convert 15 nominal features into 1 numeric one. That however is not concidered under the task of nominal to numeric conversion or vice versa. The best way to represent your data is to keep the original meanings of the features, unless you have no other choice to do so. -- Departement of Knowledge Engeneering Faculty of Humanities & Science Maastricht University http://www.netstorm.be |
|
From: Peter R. <fra...@gm...> - 2009-04-30 20:47:55
|
> In the weka explorer I want to sum several instances into a single instance > > i.e. sum each value of atrributes for a set of instances > > How to proceed ? You will have to write your own filter for that. The biggest question will be: how do you determine which instances need to be summed up? Fixed number? Common nominal attribute value (kind of a group ID)? For more information on writing a new filter, see FAQ "How do I write a new classifier or filter?". Link to the FAQs available from the Weka homepage. Cheers, Peter -- Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 |
|
From: Garykwlai <gar...@ho...> - 2009-04-30 20:27:40
|
Peter Reutemann-3 wrote: > > [...] > >> weka version 3.6.0 , JDk 1.6.0.12, Windows Vista Home Premium > > You might want to try a different maximum stack size (parameter "-Xss" > - not to be confused with "-Xmx" for the heap!). You will have to > modify your RunWeka.ini for that purpose (unless you're starting Weka > directly from the commandline using a plain "java" call). Try starting > with -Xss512k (= 512KB) and slowly increase it, if you keep getting > the problem with serialization. > > I've attached an example RunWeka.ini (can't test it, since I don't run > Windows), to replace the one in your Weka installation (never hurts to > backup the original one!). Just modify the value of the placeholder > "maxstack" (currently set to 512k). > > Cheers, Peter > -- > Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ > http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 > > > _______________________________________________ > Wekalist mailing list > Send posts to: Wekalist(a)list.scms.waikato.ac.nz > 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 > > thx it works after I set it to 1024k -- View this message in context: http://www.nabble.com/MLP-stopped-automatically-in-mid-way--tp23302392p23322609.html Sent from the WEKA mailing list archive at Nabble.com. |
|
From: Peter R. <fra...@gm...> - 2009-04-30 19:50:24
|
> But there is still a problem, the newly added attribute is named "classification" instead of the original name. so can we specify a value for the attribute name? No, this name is hard-coded in the filter (the old class attribute doesn't get deleted!). Cheers, Peter -- Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 |
|
From: ashvin <ash...@gm...> - 2009-04-30 18:36:39
|
Dear Weka users, I have a dataset with nominal and numeric attributes. I want to convert the nominal attributes to numeric. I used Nominaltobinay filter for converting my nominal attribute which has 15 distinct values and this resulted in 15 new columns in the dataset. Is there a better way of representing my data or converting the nominal to numeric type? I want to represent the nominal attribute with a single numeric value. thanks, ashvin. |
|
From: Christian S. <cs...@db...> - 2009-04-30 16:28:39
|
Hi all!
I try to mine frequent itemsets using the implementation of the Apriori-algorithm in Weka 3.6.
My rather simple example consists of 10 items and 5 (binary-encoded) itemsets:
@attribute A {1,0}
@attribute B {1,0}
@attribute D {1,0}
@attribute E {1,0}
@attribute M {1,0}
@attribute N {1,0}
@attribute O {1,0}
@attribute R {1,0}
@attribute S {1,0}
@attribute T {1,0}
@data
% E,M, O,R,S
0,0,0,1,1,0,1,1,1,0
% E, N,O, S
0,0,0,1,0,1,1,0,1,0
% A,B,D,E, N,O,R,S
1,1,1,1,0,1,1,1,1,0
% A, D, N, R,S,T
1,0,1,0,0,1,0,1,1,1
% E,M, R
0,0,0,1,1,0,0,1,0,0
As I am only interested in the frequent itemsets, I set "numRules" and "minMetric" to zero and "outputItemSets" to true.
Also I would like to see itemsets satisfying a MinSupport of at least 0.6 (= 3 Instances), so "delta" = 0, lower & higher BoundMinSupport = 0.6.
Surprisingly executing
weka.associations.Apriori -I -N 0 -T 0 -C 0.0 -D 0.0 -U 0.6 -M 0.6 -S -1.0 -V -c -1
yields
Large Itemsets L(1):
A=0 3
D=0 3
M=0 3
N=1 3
O=1 3
Which is not the complete set of frequent items, because all items with support > 0.6 (e.g. "E=1") are missing!
What am I doing wrong here?
Is it impossible to mine all frequent itemsets using Weka?
Thanks!
Christian
|
|
From: michel.plantie <mic...@em...> - 2009-04-30 14:05:30
|
Hello In the weka explorer I want to sum several instances into a single instance i.e. sum each value of atrributes for a set of instances How to proceed ? thank you for your answer MP -- ========================================================= Dr. Michel Plantié Centre de recherche LGI2P, Ecole des Mines d'Ales, Site EERIE 69, rue Georges Besse F30035 Nîmes Cedex 1 - France T.: +33 (0)466387035, fax : +33 (0)466387074 email : michel.plantie(a)ema.fr http://www.lgi2p.ema.fr/~plantie ========================================================= |
|
From: JiangJun姜军 <jia...@ic...> - 2009-04-30 11:12:39
|
Oh, a lot of thanks! This really works! But there is still a problem, the newly added attribute is named "classification" instead of the original name. so can we specify a value for the attribute name? Currently I am using the 3.6.0 stable version. > -----邮件原件----- > 发件人: wekalist-bounces(a)list.scms.waikato.ac.nz > [mailto:wekalist-bounces(a)list.scms.waikato.ac.nz] 代表 Peter Reutemann > 发送时间: 2009年4月30日 12:36 > 收件人: Weka machine learning workbench list. > 主题: Re: [Wekalist] Help: how to save predicted results back > into the arfffile? > > > Currently I am using weka for a project, I want to know > whether weka can fill the predicted values back to the arff > file. For instance, using command line: > > > > java weka.classifiers.tree.J48 -l <model-file> -T <test-file> > > > > this will appy the model to the test file and output the > relevant results, e.g. predictions, statistics, confusion > matrix. But the test-file did not change, i.e. the question > marks representing the missing value of class attribute are > not taken place by the predicted value. > > > > I wonder whether weka provides this feature, since if it > does, I only need to handle arff file, otherwise, I have to > parse the output to obtain the necessary data, which is > really more complicated. I will use C++ for development, so > modifying the source code of weka may not be a good idea. > > Read FAQ "How do I make predictions with a trained model?". The > article listed in this FAQ ("Making predictions") tells you how to use > the AddClassification filter. This filter can add predictions, class > distributions, etc to the input dataset, using a serialized model. > Link to the FAQs available from the Weka homepage. > > Cheers, Peter > -- > Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ > http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 > > _______________________________________________ > Wekalist mailing list > Send posts to: Wekalist(a)list.scms.waikato.ac.nz > 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 > |
|
From: Peter R. <fra...@gm...> - 2009-04-30 09:48:37
|
[...] > weka version 3.6.0 , JDk 1.6.0.12, Windows Vista Home Premium You might want to try a different maximum stack size (parameter "-Xss" - not to be confused with "-Xmx" for the heap!). You will have to modify your RunWeka.ini for that purpose (unless you're starting Weka directly from the commandline using a plain "java" call). Try starting with -Xss512k (= 512KB) and slowly increase it, if you keep getting the problem with serialization. I've attached an example RunWeka.ini (can't test it, since I don't run Windows), to replace the one in your Weka installation (never hurts to backup the original one!). Just modify the value of the placeholder "maxstack" (currently set to 512k). Cheers, Peter -- Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 |
|
From: Garykwlai <gar...@ho...> - 2009-04-30 09:36:58
|
Peter Reutemann-3 wrote: > > [...] > >> at >> java.io.ObjectOutputStream.writeArray(ObjectOutputStream.java:1338) >> at >> java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1146) >> >> at >> java.io.ObjectOutputStream.defaultWriteFields(ObjectOutputStream.java >> :1509) >> at >> java.io.ObjectOutputStream.writeSerialData(ObjectOutputStream.java:14 >> 74) >> at >> java.io.ObjectOutputStream.writeOrdinaryObject(ObjectOutputStream.jav >> a:1392) >> at >> java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1150) >> >> these exceptions looping over and over >> >> any idea why? > > Something with the serialization seems to fail (Weka uses Java > serialization to create copies of classifiers, etc.). > What version of Weka, JDK, OS are you using? > > Cheers, Peter > -- > Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ > http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 > > _______________________________________________ > Wekalist mailing list > Send posts to: Wekalist(a)list.scms.waikato.ac.nz > 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 > > weka version 3.6.0 , JDk 1.6.0.12, Windows Vista Home Premium -- View this message in context: http://www.nabble.com/MLP-stopped-automatically-in-mid-way--tp23302392p23312924.html Sent from the WEKA mailing list archive at Nabble.com. |
|
From: tgh <tia...@nc...> - 2009-04-30 07:23:28
|
Hi I have a data set, in which, there are 10 kinds of samples, and each kind of sample has been labeled, and each sample belong to only one kind, and I use naivebayes classifier on this data set, it seems work, I want to know whether naivebayes and updateble version in weka can classify data set into multiclass Could you help me Thanks you |
|
From: Peter R. <fra...@gm...> - 2009-04-30 04:36:25
|
> Currently I am using weka for a project, I want to know whether weka can fill the predicted values back to the arff file. For instance, using command line:
>
> java weka.classifiers.tree.J48 -l <model-file> -T <test-file>
>
> this will appy the model to the test file and output the relevant results, e.g. predictions, statistics, confusion matrix. But the test-file did not change, i.e. the question marks representing the missing value of class attribute are not taken place by the predicted value.
>
> I wonder whether weka provides this feature, since if it does, I only need to handle arff file, otherwise, I have to parse the output to obtain the necessary data, which is really more complicated. I will use C++ for development, so modifying the source code of weka may not be a good idea.
Read FAQ "How do I make predictions with a trained model?". The
article listed in this FAQ ("Making predictions") tells you how to use
the AddClassification filter. This filter can add predictions, class
distributions, etc to the input dataset, using a serialized model.
Link to the FAQs available from the Weka homepage.
Cheers, Peter
--
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174
|
|
From: JiangJun姜军 <jia...@ic...> - 2009-04-30 04:26:18
|
Hi, everyone: Currently I am using weka for a project, I want to know whether weka can fill the predicted values back to the arff file. For instance, using command line: java weka.classifiers.tree.J48 -l <model-file> -T <test-file> this will appy the model to the test file and output the relevant results, e.g. predictions, statistics, confusion matrix. But the test-file did not change, i.e. the question marks representing the missing value of class attribute are not taken place by the predicted value. I wonder whether weka provides this feature, since if it does, I only need to handle arff file, otherwise, I have to parse the output to obtain the necessary data, which is really more complicated. I will use C++ for development, so modifying the source code of weka may not be a good idea. I have sought the weka manual but failed to find it out. So I need your help! A lot of thanks! BTW: It is my first time to use mailing list, so I am not very familiar with all the etiquette. Please forgive me if I should do something wrong. |
|
From: Peter R. <fra...@gm...> - 2009-04-30 02:22:19
|
[...] > at > java.io.ObjectOutputStream.writeArray(ObjectOutputStream.java:1338) > at > java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1146) > > at > java.io.ObjectOutputStream.defaultWriteFields(ObjectOutputStream.java > :1509) > at > java.io.ObjectOutputStream.writeSerialData(ObjectOutputStream.java:14 > 74) > at > java.io.ObjectOutputStream.writeOrdinaryObject(ObjectOutputStream.jav > a:1392) > at > java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1150) > > these exceptions looping over and over > > any idea why? Something with the serialization seems to fail (Weka uses Java serialization to create copies of classifiers, etc.). What version of Weka, JDK, OS are you using? Cheers, Peter -- Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 |
|
From: Garykwlai <gar...@ho...> - 2009-04-30 02:16:58
|
Peter Reutemann-3 wrote: > >> Today I ran 2 small datasets with MLP classifier, it is very small on in >> experimenter only takes about an hour, when I run them in explorer mode, >> it >> stopped right after it created a training model. >> >> Is that suppose to happen like that? every settings are on default. > > Any exceptions in the console? > > Cheers, Peter > -- > Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ > http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 > > _______________________________________________ > Wekalist mailing list > Send posts to: Wekalist(a)list.scms.waikato.ac.nz > 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 > > at java.io.ObjectOutputStream.writeArray(ObjectOutputStream.java:1338) at java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1146) at java.io.ObjectOutputStream.defaultWriteFields(ObjectOutputStream.java :1509) at java.io.ObjectOutputStream.writeSerialData(ObjectOutputStream.java:14 74) at java.io.ObjectOutputStream.writeOrdinaryObject(ObjectOutputStream.jav a:1392) at java.io.ObjectOutputStream.writeObject0(ObjectOutputStream.java:1150) these exceptions looping over and over any idea why? -- View this message in context: http://www.nabble.com/MLP-stopped-automatically-in-mid-way--tp23302392p23306176.html Sent from the WEKA mailing list archive at Nabble.com. |
|
From: Garykwlai <gar...@ho...> - 2009-04-30 00:34:52
|
Garykwlai wrote: > > Today I ran 2 small datasets with MLP classifier, it is very small on in > experimenter only takes about an hour, when I run them in explorer mode, > it stopped right after it created a training model. > > Is that suppose to happen like that? every settings are on default. > > regards > > Gaz > oh i think i have fixed it... downloaded the weka again and reinstall seems fine now -- View this message in context: http://www.nabble.com/MLP-stopped-automatically-in-mid-way--tp23302392p23302757.html Sent from the WEKA mailing list archive at Nabble.com. |
|
From: Peter R. <fra...@gm...> - 2009-04-29 23:45:44
|
>> Sought (htdig) archive file not found >> >> ht://Dig error >> >> htsearch detected an error. Please report this to the webmaster of this >> site. The error message is: >> >> Unable to read word database file >> Did you run htmerge? > > htdig seems to be rooted... I've contacted our IT support and post > more information when I hear back from them. It's working again. Cheers, Peter -- Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 |
|
From: Peter R. <fra...@gm...> - 2009-04-29 23:29:09
|
> Sought (htdig) archive file not found > > ht://Dig error > > htsearch detected an error. Please report this to the webmaster of this > site. The error message is: > > Unable to read word database file > Did you run htmerge? htdig seems to be rooted... I've contacted our IT support and post more information when I hear back from them. Cheers, Peter -- Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 |
|
From: j f <pat...@ya...> - 2009-04-29 23:22:17
|
Sought (htdig) archive file not found ht://Dig error htsearch detected an error. Please report this to the webmaster of this site. The error message is: Unable to read word database file Did you run htmerge? thank you for this resource! kind regards, jo |
|
From: Peter R. <fra...@gm...> - 2009-04-29 21:37:33
|
> Today I ran 2 small datasets with MLP classifier, it is very small on in > experimenter only takes about an hour, when I run them in explorer mode, it > stopped right after it created a training model. > > Is that suppose to happen like that? every settings are on default. Any exceptions in the console? Cheers, Peter -- Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 |
|
From: Thomas D. <tw...@ne...> - 2009-04-29 21:25:16
|
2009/4/29 Andrew Rosenberg <amaxwell(a)cs.columbia.edu> > I think Thomas is suggesting something more like: > > BufferedReader reader = new BufferedReader(new FileReader(_fileTrain)); > ArffReader arff = new ArffReader(reader); > Instances training = arff.getData(); > training.setClassIndex(data.numAttributes() - 1); > reader.close(); // <-- these are probably not being called by weka. > > BufferedReader reader = new BufferedReader(new FileReader(_fileTest)); > ArffReader arff = new ArffReader(reader); > Instances testing = arff.getData(); > testing.setClassIndex(data.numAttributes() - 1); > reader.close(); // <-- these are probably not being called by weka. > > J48 clasificador = new J48(individuo); > clasificador.buildClassifier(training); > for (int i = 0; i < testing.numInstances(); ++i) { > Instance test_instance = testing.instance(i); > double[] distribution = > clasificador.distributionForInstance(test_instance); > > // Do something useful... > } > > > Another option is to go into the weka source and find out where those files > should be closed, and change it. My best guess is that it's in or around > ArffLoader... If you fix this, be sure to post it. > > > On Wed, Apr 29, 2009 at 2:53 PM, Thomas Debray <tw(a)netstorm.be> wrote: > >> >> >> 2009/4/29 Sandra García Rodríguez <sgrodrig(a)inf.uc3m.es> >> >>> hello, >>> >>> *if you are programming in java anyways, why not using the weka api and >>> calling the several classifiers manually? This way, you can control yourself >>> how files are handled, evaluation is done etc etc etc. >>> * >>> that's what I am doing, I call each classifier this way: >>> >>> String argv[]=new String[8]; >>> argv[0]="-t"; >>> argv[1]=_fileTrain; >>> argv[2]="-no-cv"; >>> argv[3]="-T"; >>> argv[4]=_fileTest; >>> argv[5]="-p"; >>> argv[6]="0"; >>> argv[7]="-o"; >>> J48 clasificador=new J48(individuo); >>> clasificador.runClassifier(new J48(), argv); >>> >>> I need to execute this piece of code many times in my program so, each >>> time I call J48, weka open these files but it doesn't close them, that's why >>> after a few cicles the number of open files is too big and java gives me the >>> IOException. >>> >>> Is there any way to solve it? >>> >>> Thank you very much. >>> >>> >>> >>> On Wed, Apr 29, 2009 at 4:41 PM, Thomas Debray <tw(a)netstorm.be> wrote: >>> >>>> >>>> >>>> 2009/4/29 Sandra García Rodríguez <sgrodrig(a)inf.uc3m.es> >>>> >>>>> Hello, >>>>> >>>>> I am programming something in Java which calls many many times the J48 >>>>> algorithm of weka using the parameters -t fileTrain.arff -T fileTest.arff >>>>> -no-cv -p 0 -o >>>>> >>>>> After many calls it throws this exception (it becomes from the >>>>> ArffLoader.java): >>>>> >>>>> java.io.IOException: java.io.FileNotFoundException: ./fileTrain(or >>>>> Test).arff (Too many open files) >>>>> >>>>> I think this is due to the fact that weka doesn't close the files >>>>> opened (fileTrain and fileTest) during the execution when J48 ends, is there >>>>> any way to solve that? how can I close these files when the algorithm >>>>> finish? is there any method to close all open files at the same time? >>>>> >>>>> Thank you. >>>>> >>>>> -- >>>>> Sandra García Rodríguez >>>>> Dpto. de Informática >>>>> >>>>> _______________________________________________ >>>>> Wekalist mailing list >>>>> Send posts to: Wekalist(a)list.scms.waikato.ac.nz >>>>> 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html> >>>>> >>>>> >>>> if you are programming in java anyways, why not using the weka api and >>>> calling the several classifiers manually? This way, you can control yourself >>>> how files are handled, evaluation is done etc etc etc. >>>> >>>> >>>> -- >>>> Departement of Knowledge Engeneering >>>> Faculty of Humanities & Science >>>> Maastricht University >>>> >>>> http://www.netstorm.be >>>> >>>> _______________________________________________ >>>> Wekalist mailing list >>>> Send posts to: Wekalist(a)list.scms.waikato.ac.nz >>>> 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html> >>>> >>>> >>> >>> >>> -- >>> Sandra García Rodríguez >>> Dpto. de Informática >>> >>> _______________________________________________ >>> Wekalist mailing list >>> Send posts to: Wekalist(a)list.scms.waikato.ac.nz >>> 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html> >>> >>> >> >> I actually meant calling the functions of the specific classifiers, >> rather than using their main function >> >> -- >> Departement of Knowledge Engeneering >> Faculty of Humanities & Science >> Maastricht University >> >> http://www.netstorm.be >> >> _______________________________________________ >> Wekalist mailing list >> Send posts to: Wekalist(a)list.scms.waikato.ac.nz >> 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html> >> >> > > _______________________________________________ > Wekalist mailing list > Send posts to: Wekalist(a)list.scms.waikato.ac.nz > 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html> > > I actually mean the following: DataSource trainS,testS; try { trainS = new DataSource("train.arff"); testS = new DataSource("test.arff"); Instances train = trainS.getDataSet(); Instances test = testS.getDataSet(); if (train.classIndex() == -1) train.setClassIndex(train.numAttributes() - 1); if (test.classIndex() == -1) test.setClassIndex(test.numAttributes() - 1); J48 j48 = new J48(); j48.buildClassifier(train); Evalutation eval = new Evaluation(train); eval.evaluateModel(j48,test); //close your files more info on http://weka.wiki.sourceforge.net/Use+Weka+in+your+Java+code -- Departement of Knowledge Engeneering Faculty of Humanities & Science Maastricht University http://www.netstorm.be |
|
From: Garykwlai <gar...@ho...> - 2009-04-29 21:16:50
|
Today I ran 2 small datasets with MLP classifier, it is very small on in experimenter only takes about an hour, when I run them in explorer mode, it stopped right after it created a training model. Is that suppose to happen like that? every settings are on default. regards Gaz -- View this message in context: http://www.nabble.com/MLP-stopped-automatically-in-mid-way--tp23302392p23302392.html Sent from the WEKA mailing list archive at Nabble.com. |
|
From: pufftissue <puf...@gm...> - 2009-04-29 21:04:05
|
Thank you Peter, this post absolutely helped and I would recommend others who are having problems use it! I had 74,000 instances as a reference. pufftissue wrote: > > Hi there, > > I could use some help on how to increase my heap size. I am getting an > out of memory error when trying to read a large .csv file into weka. > > I opened up terminal and typed this: java -Xmx2g > but it did not work. > > I typed in java c-classpath and it says I have none. > > I am on mac os x 10.5 leopard > > Thank you! > -- View this message in context: http://www.nabble.com/Out-of-memory-on-a-mac%2C-need-to-increase-heap-size-tp23290719p23302388.html Sent from the WEKA mailing list archive at Nabble.com. |
|
From: Peter R. <fra...@gm...> - 2009-04-29 19:49:10
|
> I have tied the “save as” option but the new version of weka 3.60 doesn’t > have the option to save as LIBSVM. > > How is it possible to convert a comma separated value dataset to LIBSVM > format using WEKA? I assume that you're talking about the "save as" option in the Explorer. Not all savers support all attribute types. The LibSVM saver only supports nominal and numeric attributes (and needs a class attribute set). If your dataset contains string attributes or has no class attribute set, you won't be able to save it with this saver. Cheers, Peter -- Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174 |
|
From: Andrew R. <ama...@cs...> - 2009-04-29 19:13:54
|
I think Thomas is suggesting something more like:
BufferedReader reader = new BufferedReader(new FileReader(_fileTrain));
ArffReader arff = new ArffReader(reader);
Instances training = arff.getData();
training.setClassIndex(data.numAttributes() - 1);
reader.close(); // <-- these are probably not being called by weka.
BufferedReader reader = new BufferedReader(new FileReader(_fileTest));
ArffReader arff = new ArffReader(reader);
Instances testing = arff.getData();
testing.setClassIndex(data.numAttributes() - 1);
reader.close(); // <-- these are probably not being called by weka.
J48 clasificador = new J48(individuo);
clasificador.buildClassifier(training);
for (int i = 0; i < testing.numInstances(); ++i) {
Instance test_instance = testing.instance(i);
double[] distribution =
clasificador.distributionForInstance(test_instance);
// Do something useful...
}
Another option is to go into the weka source and find out where those files
should be closed, and change it. My best guess is that it's in or around
ArffLoader... If you fix this, be sure to post it.
On Wed, Apr 29, 2009 at 2:53 PM, Thomas Debray <tw(a)netstorm.be> wrote:
>
>
> 2009/4/29 Sandra García Rodríguez <sgrodrig(a)inf.uc3m.es>
>
>> hello,
>>
>> *if you are programming in java anyways, why not using the weka api and
>> calling the several classifiers manually? This way, you can control yourself
>> how files are handled, evaluation is done etc etc etc.
>> *
>> that's what I am doing, I call each classifier this way:
>>
>> String argv[]=new String[8];
>> argv[0]="-t";
>> argv[1]=_fileTrain;
>> argv[2]="-no-cv";
>> argv[3]="-T";
>> argv[4]=_fileTest;
>> argv[5]="-p";
>> argv[6]="0";
>> argv[7]="-o";
>> J48 clasificador=new J48(individuo);
>> clasificador.runClassifier(new J48(), argv);
>>
>> I need to execute this piece of code many times in my program so, each
>> time I call J48, weka open these files but it doesn't close them, that's why
>> after a few cicles the number of open files is too big and java gives me the
>> IOException.
>>
>> Is there any way to solve it?
>>
>> Thank you very much.
>>
>>
>>
>> On Wed, Apr 29, 2009 at 4:41 PM, Thomas Debray <tw(a)netstorm.be> wrote:
>>
>>>
>>>
>>> 2009/4/29 Sandra García Rodríguez <sgrodrig(a)inf.uc3m.es>
>>>
>>>> Hello,
>>>>
>>>> I am programming something in Java which calls many many times the J48
>>>> algorithm of weka using the parameters -t fileTrain.arff -T fileTest.arff
>>>> -no-cv -p 0 -o
>>>>
>>>> After many calls it throws this exception (it becomes from the
>>>> ArffLoader.java):
>>>>
>>>> java.io.IOException: java.io.FileNotFoundException: ./fileTrain(or
>>>> Test).arff (Too many open files)
>>>>
>>>> I think this is due to the fact that weka doesn't close the files opened
>>>> (fileTrain and fileTest) during the execution when J48 ends, is there any
>>>> way to solve that? how can I close these files when the algorithm finish? is
>>>> there any method to close all open files at the same time?
>>>>
>>>> Thank you.
>>>>
>>>> --
>>>> Sandra García Rodríguez
>>>> Dpto. de Informática
>>>>
>>>> _______________________________________________
>>>> Wekalist mailing list
>>>> Send posts to: Wekalist(a)list.scms.waikato.ac.nz
>>>> 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html>
>>>>
>>>>
>>> if you are programming in java anyways, why not using the weka api and
>>> calling the several classifiers manually? This way, you can control yourself
>>> how files are handled, evaluation is done etc etc etc.
>>>
>>>
>>> --
>>> Departement of Knowledge Engeneering
>>> Faculty of Humanities & Science
>>> Maastricht University
>>>
>>> http://www.netstorm.be
>>>
>>> _______________________________________________
>>> Wekalist mailing list
>>> Send posts to: Wekalist(a)list.scms.waikato.ac.nz
>>> 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html>
>>>
>>>
>>
>>
>> --
>> Sandra García Rodríguez
>> Dpto. de Informática
>>
>> _______________________________________________
>> Wekalist mailing list
>> Send posts to: Wekalist(a)list.scms.waikato.ac.nz
>> 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html>
>>
>>
>
> I actually meant calling the functions of the specific classifiers, rather
> than using their main function
>
> --
> Departement of Knowledge Engeneering
> Faculty of Humanities & Science
> Maastricht University
>
> http://www.netstorm.be
>
> _______________________________________________
> Wekalist mailing list
> Send posts to: Wekalist(a)list.scms.waikato.ac.nz
> 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<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html>
>
>
|