mi-erasmusmc / Explore

Finding a short and accurate decision rule in disjunctive normal form by exhaustive search
https://mi-erasmusmc.github.io/Explore/
1 stars 1 forks source link

Problems finding best length using cross validation in parallel #22

Open AniekMarkus opened 3 months ago

AniekMarkus commented 3 months ago

Running EXPLORE for mix_4.arff with settings:

_PartitionMethod=CROSSVALIDATION StartRulelength=1 EndRulelength=5 LearnRatio=0.7 NumberofPartitions=3 Subsumption=no BranchBound=no Parallel=yes (fo 'no' EXPLORE successfully completes) ParallelMethod=ONE BinaryReduction=yes

Sometimes gives error messages after 1 or 2 partitions (e.g. the one below) and sometimes successfully completes: libc++abi: terminating due to uncaught exception of type std::length_error: vector Process finished with exit code 134 (interrupted by signal 6: SIGABRT)

In addition: when re-training model with full train set (learn + validate), the functions induce() etc. not yet available in parallel.