Tony Bagnall

this is so out of date we should close it

I agree with this, I think you will just run into the inherent instability of TensorFlow, and the utility of these tests imo vanishes if you have to engineer it...

you can also load the regression and forecasting and regression archive data :)

if you focus on the 112, I can do the new ones. Are you running with sktime default parameter?

yeah we should check that, I should have thought about that!

currently recognises all 19 multivariate capable ones (in version 0.12). these are 'Arsenal', 'CNNClassifier', 'CanonicalIntervalForest', 'Catch22Classifier', 'ColumnEnsembleClassifier', 'DrCIF', 'FreshPRINCE', 'HIVECOTEV2', 'IndividualTDE', 'KNeighborsTimeSeriesClassifier', 'MUSE', 'ProbabilityThresholdEarlyClassifier', 'RandomIntervalClassifier', 'RocketClassifier', 'ShapeletTransformClassifier', 'SignatureClassifier', 'SummaryClassifier', 'TSFreshClassifier',...

there are 37 classifiers in total, so 18 are univariate only. 18 Univariate only ['BOSSEnsemble', 'ClassifierPipeline', 'ComposableTimeSeriesForestClassifier', 'ContractableBOSS', 'DummyClassifier', 'ElasticEnsemble', 'HIVECOTEV1', 'IndividualBOSS', 'MatrixProfileClassifier', 'ProximityForest', 'ProximityStump', 'ProximityTree', 'RandomIntervalSpectralEnsemble', 'ShapeDTW', 'SklearnClassifierPipeline', 'SupervisedTimeSeriesForest',...

First run through problems: DrCIF crashes on one of the new MTSC datasets, EmoPain, that does not occur with the other three base classifiers. It looks like an interval is...

all (STC, DrCIF, STC, TDE) run out of memory on InsectWingbeatEq, re run with 64GB

do not have equal length JapaneseVowelEq problem