def test_all_chains():
nomix =[True, False]
clf_type = ['svm', 'rf', None]
exact = [1, 2, 3, 4]
threshold = [1, 2, 3, 4, 5, 6]
results = dict()
for c in clf_type:
for m in nomix:
for t in threshold:
results[('threshold', c, m, t)] = predict_threshold(clf_type=c, nomix=m, threshold=t)
for m in nomix:
for t in threshold:
results[('duo_threshold', c, m, t)] = predict_duo_threshold(nomix=m, threshold=t)
for m in nomix:
results[('duo', c, m)] = predict_duo(nomix=m)
for c in clf_type:
for m in nomix:
for t in threshold:
for e in exact:
results[('exact_threshold', c, m, t, e)] = predict_exact_or_threshold(nomix=m, clf_type=c, threshold=t, exact=e)
threshold = [1, 2]
for t in threshold:
results[('onlymix_threshold', t)] = predict_threshold(nomix=False, only_mix=True, threshold=t)
data.serialize_keep_copy(settings.ANALYSIS_FOLDER + '/chaining.pkl', results)
return results
def _serialize_cv_results(stats_list, persist, where):
res_stats_map = defaultdict(list)
for s, data_set_id in stats_list:
res_stats_map[data_set_id].append(s)
if persist:
data.serialize_keep_copy(where, res_stats_map, keep_copy=True)
return res_stats_map
def serialize(self, where=None, keep_copy=True):
if not where:
stats_file = settings.EVAL_FOLDER + self.id
else:
stats_file = where + self.id
data.serialize_keep_copy(stats_file, self, keep_copy=keep_copy)
def serialize(self, to=settings.CLF_FOLDER, id=''):
"""
Write classifier to disk.
:param to: folder where to save
:return:
"""
to += '/{!s}_{!s}_{!s}.pkl'.format(self.dga, self.clf_type, id)
data.serialize_keep_copy(to, self)
def test_rf_mix_chains():
results = dict()
threshold = [1, 2]
clfs_all = classifiers.ClassificationEnsemble(nomix=False, only_type='rf', only_mix=True).clfs
for i in range(3, 19):
for t in threshold:
clfs_sel = classifiers.ClassificationEnsemble(nomix=False, only_type='rf', only_mix=True,clf_list=random.sample(clfs_all, i))
results[('mix_rf', t, i)] = predict_threshold(nomix=False, only_mix=True, threshold=t, clfs=clfs_sel)
data.serialize_keep_copy(settings.ANALYSIS_FOLDER + '/chaining.pkl', results)
return results
def predict_all_mixed_sets_on_x(n_jobs=8):
w = Workspace(days=1, empty=True)
w.load_all(settings.SetTypes.mixed_dga_grouped_family.value)
clfs = classifiers.ClassificationEnsemble().clfs
clfs = [c for c in clfs if 'mix' not in c.dga_type]
parallel = Parallel(n_jobs=n_jobs, verbose=1)
ds = w.data_sets_loaded['mixed_dga_grouped_family_50000_59_2.pkl']
dmns, lbls, groups = ds.expand()
# res is list of tuples: clf.clf_type, clf.dga_type, lbls, pred_lbl
res = parallel(
delayed(eval_train_test.predict_all_on_x)(clf, ds, dmns, lbls, groups)
for clf in clfs
)
data.serialize_keep_copy(settings.ANALYSIS_FOLDER + '/x_vs_all_results.pkl', (res, dmns))