Activity for Jesse Read
-
Jesse Read posted a comment on discussion General Discussion
Hi, Probably your data is multi-target and not multi-label so it is only using the multi-target metrics. As the target variables are thus not binary, only results for a restricted set of metrics can be produced. Although it could be argued that Precision, Recall and F1 scores may be applied also in multi-target evaluation, this is not currently implemented in Meka. Jesse
-
Unknown Values in Output Vectors Become 0 When Evaluating
-
update for new release
-
MEKA CODE REPOSITORY HAS NOW MOVED TO: https://...
-
Final update to README, version number
-
Markdown fixed in README
-
Delete README.txt (has been replaced by README.md)
-
README updated; changed from txt to markdown fo...
-
Renamed Show*PrecisionRecall to, more generical...
-
Minor update to tutorial
-
New ref files from tests
-
Update .ref files from tests
-
Tidy up SCC and supporting code
-
Updated version number, prepared for next release
-
Updated tutorial, and included new screenshots.
-
Reorganized tutorial (section 4); added Explore...
-
Updated README with recent developments
-
Added threshold data and visualization plugin f...
-
Added Macro-Average PRC, and Macro and Micro Pr...
-
Added flatten() method for double[][] matrices ...
-
Added a visualization plugin to show micro-aver...
-
Added Micro-averaged Curve data to Results
-
Added method for getting a micro-averaged curve
-
Added 'average rank' row
-
More efficient implementation of some Metrics
-
Mentioned that high -verbosity will cause extra...
-
Bug fix for obtaining ranking matrix.
-
CT prepares a default chain in the same way as CC
-
Added methods to obtain Value and Rank matrices...
-
Added method for Nemenyi test (not yet implemen...
-
Fixed bug in log statement: folds numbered from...
-
Nicer production of Latex output
-
Added multitarget version of RAkELd, which shar...
-
Fixed RAkELd, wrt earlier modifications to PSUtils
-
SCC now extends NSR, makes more use of existing...
-
update to NSR
-
Refactor some super-label utils, and moving som...
-
Fixed bug where attempted to apply threshold wh...
-
Latest improvements mentioned in README
-
Result class now contains getAvailableMetrics(....
-
Finished SaveCSV plugin for incremental evaluat...
-
Training dataset should not be required again i...
-
Using setMeasurement(String,Object), getMeasure...
-
Using setMeasurement(String,Object),getMeasurem...
-
Plugin to save output as CSV file
-
Bug fix regarding getK() method
-
Bug fix in table output
-
RakEL now extends RakELd, and not the other awy...
-
Nicer Latex output from experiments.
-
Add Latex class for Latex output frome experime...
-
Add LatexMeasurement class for providing latex ...
-
Variables linked to options should not be chang...
-
CDT now interprets '0' width like CT.
-
For CT, allow width of 0 to indicate width of L...
-
Added Levenshtein distance to multi-target metr...
-
Update tutorial explaining prequential vs batch...
-
Cleanup of Results class; renaming/adding methods.
-
Samples option is also Batches option for batch...
-
Restored and fixed 'batch-incremental' evaluation.
-
Editions to tutorial
-
Added old batch-incremental option for incremen...
-
Tutorial updated regarding recent changes
-
Command-line options fixed, change to 'supervis...
-
Updated README with recent changes
-
Improve evaluation command-line help text
-
More revision of the Tutorial
-
New material, highlighting points which need at...
-
Incremental samples now include #Instances, #Sa...
-
Add note about averaging when type endsWith 'CV'
-
DeepML takes a MultiLabel base classifier
-
Fixed MCC 'inferenceIterations' option and tooltip
-
Update tutorial
-
update javadoc comments and outline future dev ...
-
MajorityLabelset is multi-target capable.
-
Restore multi-target incremental eval; display ...
-
Fixed bug in BPNN (Random number generator not ...
-
Added flag to allow for more memory to java
-
Fixed bug causing exception on labelsets.
-
Test/train time calculated correctly (cumulativ...
-
Option for number-of-samples becomes -x
-
PrequentialEvaluation now includes Instances wi...
-
'number of folds' determines number of samples ...
-
code cleanup
-
set/get methods and tool tips added/cleaned up;...
-
Added tooltips, cleaned up options code
-
AbstractNeuralNet implements Randomizable
-
Fixed tip text methods
-
Added set/get methods, tip text
-
made note of recent changes
-
A fresh initial look at the experimenter
-
Cleaned up options and tool tips
-
Fixed crash where empy labelset appears in dist...
-
Introduce AbstractMultiLabelClassifier, e.g., M...
-
setUnlabelledData renamed introduceUnlabelledDa...
-
setChain(..) renamed to prepareChain(..)
-
Cleaned up false javadoc flags
-
Payoff option (-P) set by String and uses any M...
-
MCC now uses CC's prepareChain method.
-
Update log of recent changes
-
Removed debug output
1 >