def test_model(gm):
full_csv_path = os.path.join(ARTIFACT_DIR, TEST_INFO_FILE)
result = []
with closing(open(full_csv_path, 'r')) as csvfile:
dr = DictReader(csvfile)
for row in dr:
full_sample_path = os.path.join(ARTIFACT_DIR, row['filename'])
feat = analyze.extract_features_file(full_sample_path)
score = gm.score(feat)
is_ok = row['is_ok'] == '1'
result.append(TestResult(row['filename'], score, is_ok))
return result
def generate_features():
full_csv_path = os.path.join(ARTIFACT_DIR, TRAINING_INFO_FILE)
feat_stacked = None
with closing(open(full_csv_path, 'r')) as csvfile:
dr = DictReader(csvfile)
for row in dr:
sample_path = os.path.join(ARTIFACT_DIR, row['filename'])
feat = analyze.extract_features_file(sample_path)
if feat_stacked is None:
feat_stacked = feat
else:
feat_stacked = np.vstack((feat_stacked, feat))
return feat_stacked
def create_features_data(filename):
feat_finale = None
labels_finale = None
with closing(open(filename, 'r')) as csvfile:
dr = DictReader(csvfile)
for row in dr:
feat = analyze.extract_features_file(row['filename'])
labels = np.ones((feat.shape[0], 1)) * float(row['is_ok'])
if feat_finale is None:
feat_finale = feat
else:
feat_finale = np.vstack((feat_finale, feat))
if labels_finale is None:
labels_finale = labels
else:
labels_finale = np.vstack((labels_finale, labels))
return feat_finale, labels_finale
def determine_from_sample(filename, clsfr):
feat = analyze.extract_features_file(filename)
return determine(clsfr, feat)
def sample_test(gm, sample_filename):
feat = analyze.extract_features_file(sample_filename)
return gm.score(feat)