10
folds and 1
repetitions.The performance is measured in three different ways.
algo | model | succ | par10 | mcp |
---|---|---|---|---|
baseline | vbs | 0.769 | 8377.272 | 0.000 |
baseline | singleBest | 0.677 | 11703.780 | 324.411 |
baseline | singleBestByPar | 0.677 | 11703.780 | 324.411 |
baseline | singleBestBySuccesses | 0.677 | 11703.780 | 324.411 |
regr | lm | 0.675 | 12135.062 | 324.411 |
regr | rpart | 0.628 | 13900.395 | 538.191 |
regr | randomForest | 0.717 | 10681.351 | 190.026 |
The following default feature steps were used for model building:
ALL, code, AST
Number of presolved instances: 0
The cost for using the feature steps (adapted for presolving) is: 0
or on average: 0
The feature steps correspond to the following 54
/ 54
instance features:
nvarsOrig, nclausesOrig, nvars, nclauses, reducedVars,
reducedClauses, Pre.featuretime, vars.clauses.ratio, POSNEG.RATIO.CLAUSE.mean, POSNEG.RATIO.CLAUSE.coeff.variation,
POSNEG.RATIO.CLAUSE.min, POSNEG.RATIO.CLAUSE.max, POSNEG.RATIO.CLAUSE.entropy, VCG.CLAUSE.mean, VCG.CLAUSE.coeff.variation,
VCG.CLAUSE.min, VCG.CLAUSE.max, VCG.CLAUSE.entropy, UNARY, BINARY.,
TRINARY., Basic.featuretime, VCG.VAR.mean, VCG.VAR.coeff.variation, VCG.VAR.min,
VCG.VAR.max, VCG.VAR.entropy, POSNEG.RATIO.VAR.mean, POSNEG.RATIO.VAR.stdev, POSNEG.RATIO.VAR.min,
POSNEG.RATIO.VAR.max, POSNEG.RATIO.VAR.entropy, HORNY.VAR.mean, HORNY.VAR.coeff.variation, HORNY.VAR.min,
HORNY.VAR.max, HORNY.VAR.entropy, horn.clauses.fraction, VG.mean, VG.coeff.variation,
VG.min, VG.max, KLB.featuretime, CG.mean, CG.coeff.variation,
CG.min, CG.max, CG.entropy, cluster.coeff.mean, cluster.coeff.coeff.variation,
cluster.coeff.min, cluster.coeff.max, cluster.coeff.entropy, CG.featuretime
The feature steps correspond to the following 75
/ 75
algorithm features:
Lines..Average., Lines..Total., Size..Average., Size..Total., Number.of.files,
Cyclomatic..Average., Cyclomatic..Total., Max.Indent..Average., Max.Indent..Total., nb_nodes,
nb_edges, degree_min, degree_max, degree_mean, degree_variance,
degree_entropy, transitivity, clustering_min, clustering_max, clustering_mean,
clustering_variance, path_min, paths_max, path_mean, path_variance,
path_entropy, Stmt, Type, Decl, Attribute,
Operator, Literal, edge_ss, edge_st, edge_sd,
edge_sa, edge_so, edge_sl, edge_ts, edge_tt,
edge_td, edge_ta, edge_to, edge_tl, edge_ds,
edge_dt, edge_dd, edge_da, edge_do, edge_dl,
edge_as, edge_at, edge_ad, edge_aa, edge_ao,
edge_al, edge_os, edge_ot, edge_od, edge_oa,
edge_oo, edge_ol, edge_ls, edge_lt, edge_ld,
edge_la, edge_lo, edge_ll, op_short, op_int,
op_long, op_long_long, op_float, op_double, op_bit
7
to 2
, resulting in a PAR10 score of 10667.522
for the reduced model. Analogously, the model that was generated based on 6
of the originally 109
features resulted in a PAR10 score of 10631.329
.
Below, you can find the list of the selected features and solvers:
Selected Features:
nclausesOrig, VG.mean, VG.min, Number.of.files, Cyclomatic..Average.,
op_float
Selected Solvers:
MaxHS, UWrMaxSAT