naywhayare changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
sumedhghaisas has joined #mlpack
< sumedhghaisas> naywhayare: Hey ryan, had some doubts about abstractions, you free??
govg has quit [Ping timeout: 240 seconds]
udit_s has joined #mlpack
sumedhghaisas has quit [Ping timeout: 240 seconds]
Anand has joined #mlpack
< Anand> Marcus : Did you have a look at the tests?
Anand has quit [Ping timeout: 240 seconds]
Anand has joined #mlpack
udit_s has quit [Read error: Connection reset by peer]
< marcus_zoq> Anand: Sorry for the slow reponse, the test is looking good.
< Anand> Marcus : I just sent you a mail. Please check!
< marcus_zoq> Anand: Maybe it would be good if you start with the nbc method, to integrate the metric not only for the mlpack nbc method, but also for the other libraries: shogun, scikit, etc. So that we can comprare the results.
udit_s has joined #mlpack
< marcus_zoq> Anand: The intergration should be the same for the other methods.
< Anand> Ok. That is a good idea!
< Anand> No not exactly!
< Anand> Because not all other methods are exactly binary/multiclass classifiers
< marcus_zoq> Anand: Yeah, you are right!
< Anand> We will need to think about that a bit
< Anand> Presently, I will work on integrating the metrics for NBC method in all other libraries too
< Anand> Do we have documentation for their code?
< Anand> I will start looking at their code
< marcus_zoq> Anand: Okay, good. All libraries 'have' a documentation. A good basis should be the existing code, we just need to get the right values.
< Anand> Ok. To get to correct integration, I have to go through their code and documentation
< marcus_zoq> Anand: Yeah, so for the scikit method I think you can use the 'predic_proba' method to get the probability for the test values.
< marcus_zoq> Anand: Here is the link: http://scikit-learn.org/stable/modules/generated/sklearn.naive_bayes.BernoulliNB.html#sklearn.naive_bayes.BernoulliNB
< Anand> Ok. Thanks. I will have a look.
< marcus_zoq> Anand: For the weka method you can use the 'distributionForInstance(Instance instance)' function: http://weka.sourceforge.net/doc.dev/weka/classifiers/bayes/NaiveBayes.html.
< marcus_zoq> Anand: All weka methods are written in java.
< Anand> I can call all those methods just like we call mlpack NBC in nbc.py
< Anand> right?
< marcus_zoq> To get the timing results, it's the same way. For the mlpack method we use the nbc binary and for the shogun and scikit method we use the python interface to run the method. For all weka method I've written a java method to get the results and in the weka nbc.py method I use this java class.
< marcus_zoq> Anand: But all methods implements the same interface. So you can integrate the metrics in the same way.
udit_s has quit [Quit: Leaving]
Anand has quit [Ping timeout: 240 seconds]
udit_s has joined #mlpack
udit_s has quit [Client Quit]
sumedhghaisas has joined #mlpack
oldbeardo has joined #mlpack
oldbeardo has quit [Changing host]
oldbeardo has joined #mlpack
oldbeardo has quit [Quit: Page closed]
udit_s has joined #mlpack
sumedhghaisas has quit [Ping timeout: 240 seconds]
sumedhghaisas has joined #mlpack
< sumedhghaisas> Hey Ryan, free for some time??
< sumedhghaisas> naywhayare: sorry forgot to mention.....
udit_s has quit [Quit: Leaving]
sumedhghaisas has quit [Ping timeout: 245 seconds]
witness___ has quit [Quit: Connection closed for inactivity]
sumedhghaisas has joined #mlpack