I am using PCA for extracting features of a training set and then
use for classification.
I have a question:
How to extratct features of a test set via PCA obtained with a training set?
Although the question of getting Weka results into mySQL has been
answered (if I'm correct), I can't find information on how to convert the
mySQL information to an arff file - the datasets (.sql file) are quite
large. I am using Weka 3.5.
Btw I am new to Weka so I am not really experienced in using this.
I have a question about leave-one-out cross validation.
My understanding about LOOCV is that one case is left to be testing case
while the rest of the dataset are the training cases.
The number of running training process is equal to the number of cases
in the dataset.
So even when I change the seed, the result should be the same.
However, the accuracy is different when I try different seeds.
Btw, I changed the seed via the GUI in explorer: "Random seed for XVal /
Thank you in advance for your time,
IMPORTANT WARNING: This email (and any attachments) is only intended for the use of the person or entity to which it is addressed, and may contain information that is privileged and confidential. You, the recipient, are obligated to maintain it in a safe, secure and confidential manner. Unauthorized redisclosure or failure to maintain confidentiality may subject you to federal and state penalties. If you are not the intended recipient, please immediately notify us by return email, and delete this message from your computer.
I need to do data mining with weka, but the results are not satisfactory with the cross validation, can someone help me?
Sur Windows Live Ideas, découvrez en exclusivité de nouveaux services en ligne... si nouveaux qu'ils ne sont pas encore sortis officiellement sur le marché !
dear weka forum,
I had gone through all the documentations present in source forge, but i could not find clear explanation of the algorithms.Can some one tell me where i can find them in Internet.
Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now.
Hello. I have a csv file whose contents I will paste in below. It is culled
from tweets from twitter and thus has a lot of messy characters. I am having
trouble bringing it into Weka. I get error messages like
... not recognized as a 'CSV' data files' file
wrong number of values. Read 9, expected 8, read Token[EOL], line 10.
I am not sure what kind of preprocessing I could do to make this data
acceptable to Weka. Can anyone suggest what needs to be done?
Thank you. Here is the file (there should be 8 fields):
30758980,10:51:11,4,12,2009,elisanajera,An empty stomach is not a good
political adviser: Einstein.,Everywhere
30758981,10:51:11,4,12,2009,Phaetonchix,is doing research for her upcoming
30758982,10:51:11,4,12,2009,zombieomatic,ALERT: ZOMBIES GONE WILD a lego
30758999,10:51:12,4,12,2009,KStewsPussy,"@_Team_Robsten Yeah I"ll tell
her that..But I wnated to be sick",somewhere near KStewsButt
30759000,10:51:12,4,12,2009,generatorx,"Mark Napier at [DAM]Berlin
tomorrow deconstructing Pam Anderson as a modern-day Venus
30759010,10:51:13,4,12,2009,SAMM_wich,RT @wiillz What up doe- imy :(,new
30759026,10:51:14,4,12,2009,SacTownRadio,Follow these good people - @othtv
@CHUCC1THACHIEF @GoldenMeanSteph @beto916 @TAJMACSWAMPZENT @TheRealBueno
@mrsanncarter @MDash707,NORTHERN CALI
30759027,10:51:14,4,12,2009,DaHeartlessLova,I"m so fucking dumb I want
diamonds on my thumb. Got diamonds on my dick so that"s diamonds on her
tongue. Brrrr lol,The City Of Devils
30759031,10:51:14,4,12,2009,smalldogs,who honestly cares about Tiger Woods
and his pants party?,long beach
30759069,10:51:17,4,12,2009,VRWCTexan,@j_marie_a1 As a direct decendent of
brave souls that fought at Battle of Goliad & signed TX Declaration of
Independance - happy to bestow!,Greater Houston Area
30759096,10:51:19,4,12,2009,sunshine_alina,"just voted ""Rihanna & Chris
Brown"" on ""The 2009 breakup you will never forget"" vote too ➔
30759158,10:51:23,4,12,2009,jarehoppipola,My Twitter account is worth $84!
What"s yours worth? http://bit.ly/sFMkd,Bandung
30759173,10:51:25,4,12,2009,Net_News_Global,NNG: American Sponsorship of
Global Terrorism http://bit.ly/8IUobt,Germany
30759194,10:51:27,4,12,2009,iyaDedE,So ladies let me tell you what"s cool
this winter! Ski suits + war make up. Eyelids or lips is not the only place
u can make up! Hint hint!,New York+Rwanda+Belgium
I want to do regression via LibSVM. I know we have following options in SVM
choice. So as long as I guess, 3-4 are for regression.
0 = C-SVC
1 = nu-SVC
2 = one-class SVM
3 = epsilon-SVR
4 = nu-SVR
But still It gives error:
"weka.classifiers.functions.LibSVM: Cannot handle numeric class!"
View this message in context: http://old.nabble.com/Regression-via-SVM-tp26977796p26977796.html
Sent from the WEKA mailing list archive at Nabble.com.
Based on the messages posted in the mainling list, I understand that
increasing heap size can be either by changing the RunWeka.bat batch file
(if running weka via start menu) or by using -Xmx256m from the command line.
Am using weka 3.7.0 and NetBeans IDE 6.7.1 for compiling and running weka.
Both solutions listed above can not be used in my case,
and i can't find anything under project configurations (such as "VM
Arguments") which allow me to specify heap size.
Finally, i have tried increasing the heap size by editing the "*
netbeans.conf*" file, in which i changed the option *-J-Xmx32m* to *
-J-Xmx256m*.. Then when i run weka again and check (using *java
weka.core.SystemInfo*) to check heap size, it lists:
memory.initial: 4.9MB (5177344)
memory.max: 63.6MB (66650112)
I can not understand why it's not increased?!!
For those using NetBeans for running Weka, could you please provide other
suggestions or correct me if my understanding is wrong.