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/
andrewmw94 has left #mlpack []
sumedh__ has joined #mlpack
Anand has joined #mlpack
< Anand>
Marcus : Two things
udit has joined #mlpack
Anand_ has joined #mlpack
< Anand_>
Marcus : Two things - 1. I don't think the weka linear regression works even with the arff format. 2. The parameters.csv in mlpack linear regression doesn't seem to be the correct file for predictions.
Anand has quit [Ping timeout: 246 seconds]
Anand_ has quit [Ping timeout: 246 seconds]
< sumedh__>
naywhayare: The tests are complete... added ConvergeTest... but how to tell that the algorithm is converging... ??
< sumedh__>
I always remember Turing's halting problem when I refer to this convergence problem :)
< sumedh__>
anyways... we need to return information about the termination type with amf Apply...
< sumedh__>
what I mean is... we need to return if the algorithm converged or reached maxIterations...
< sumedh__>
what do you think??
udit has quit [Quit: leaving]
andrewmw94 has joined #mlpack
< naywhayare>
sumedh__: could you test the residue in the test and make sure it is small?
< naywhayare>
or the validation RMSE?
sumedh_ has joined #mlpack
sumedh__ has quit [Ping timeout: 240 seconds]
Anand has joined #mlpack
< Anand>
Marcus : I dont think that a predictions.csv is generated by mlpack linear regression
< Anand>
Ok no there is a predictions.csv. Just saw it in the mlpack code
< Anand>
Marcus : Tell me whenever you are free.
< marcus_zoq>
Anand: So that solves the second problem?
< Anand>
Can you check once if it is working for you/
< Anand>
?
< Anand>
I dont know why but Iam not able to run it. Error -> [FATAL] Could not execute command: ['linear_regression', '-i', 'datasets/iris_train.csv', '-t', 'datasets/iris_test.csv', '-v']
< marcus_zoq>
Anand: Can you run 'linear_regression -i datasets/iris_train.csv -t datasets/iris_test.csv -v' from the repo root?
< Anand>
error: Mat::init(): requested size is not compatible with column vector layout terminate called after throwing an instance of 'std::logic_error' what(): Mat::init(): requested size is not compatible with column vector layout Aborted (core dumped)
< Anand>
Are you able to run it?
< marcus_zoq>
Anand: No, because we are trying to transpose a colvector ...
< marcus_zoq>
Anand: Or we are trying to save a rowvector into a colvector
< Anand>
Marcus : Where are we possibly doing this?
< marcus_zoq>
Anand: Line: 173 in the linear_regression_main.cpp file.
< Anand>
Ok. So should we remove the transpose?
< marcus_zoq>
Anand: Yes, we need to remove this line and write data::Save(outputPredictions, predictions, true, false);.
< Anand>
That will require a new build.
< Anand>
May be you should include this in the new release too.
< marcus_zoq>
Anand: Yeah, good point.
< marcus_zoq>
Anand: Do you use the svn repository?
< Anand>
Yes, I built the library from svn
< marcus_zoq>
Anand: Okay good.
< Anand>
This should solve out second problem. What about weka?
< marcus_zoq>
Anand: I had not yet had time to look into the issue. Any ideas?
< Anand>
Marcus : Not sure, I am getting some parser errors
< Anand>
And some errors like these:
< Anand>
ValueError: Some errors were detected ! Line #6 (got 2 columns instead of 1) Line #10 (got 2 columns instead of 1) Line #11 (got 4 columns instead of 1) Line #12 (got 3 columns instead of 1)
< Anand>
And the files are not getting generated
< marcus_zoq>
Anand: Can you insert an print(s) after s = subprocess.check_output(.. in line 80?
< marcus_zoq>
Anand: And past the output?
< marcus_zoq>
paste
< Anand>
b"---Registering Weka Editors---\nTrying to add database driver (JDBC): RmiJdbc.RJDriver - Error, not in CLASSPATH?\nTrying to add database driver (JDBC): jdbc.idbDriver - Error, not in CLASSPATH?\nTrying to add database driver (JDBC): org.gjt.mm.mysql.Driver - Error, not in CLASSPATH?\nTrying to add database driver (JDBC): com.mckoi.JDBCDriver - Error, not in CLASSPATH?\nTrying to add database driver (JDBC): org.hsqldb.jdbcDriver -
< Anand>
Some random output?
< marcus_zoq>
Anand: Okay the first thing we need to change is line 81 use input_responsesFile instead of regressorsFile.
Anand_ has joined #mlpack
< marcus_zoq>
But this does not solve the problem. Only takes us to the next problem ...
< Anand_>
Line 81?
< marcus_zoq>
In the LinearRegression.java file.
Anand has quit [Ping timeout: 246 seconds]
< marcus_zoq>
'if (regressorsFile.length() != 0)'
< Anand_>
Yeah, still errors
< marcus_zoq>
Anand_: The same errors?
< Anand_>
Yes
< marcus_zoq>
Anand_: Okay, I thought we get another error from the java.
< Anand_>
Are you getting some idea about this error? I still am not able to.
< jenkins-mlpack>
Starting build #2013 for job mlpack - svn checkin test (previous build: SUCCESS)
< marcus_zoq>
Anand_: Not really, I've just looked at the code, I think I should use a debugger. Right now I'm busy, maybe that can wait until tomorrow?
< Anand_>
Ok. The merge has to wait more?
< Anand_>
I guess so
< marcus_zoq>
Anand_: Yeah, maybe you can apply the code guidelines so that we can merge as soon we fixed the error
< Anand_>
Sure, I will do it now, wherever I find a need
Anand_ has quit [Ping timeout: 246 seconds]
< sumedh_>
naywhayare: but user should be able to get this information somehow...
< sumedh_>
right now there is no way to know how the algorithm terminated...