pkpdai / pkdocclassifier Goto Github PK

Issue #1:

Running section 3.1.3, README.md

python scripts/bootstrap_bow.py \
   --input-dir data/encoded/ngrams \
   --output-dir data/results/ngrams \
   --output-dir-bootstrap data/results/ngrams/bootstrap \
   --path-labels data/labels/dev_data.csv

causes AssertionError from pk_classifier/bootstrap.py, line 95:

assert features['pmid'].equals(labs['pmid'].astype('int32'))

Cause: inappropriate --path-labels input
Proposed change: dev_data.csv--> test_data.csv (README, section 3.1.3)

Issue #2:

After fixing Issue #1, the same error occurred. Upon closer inspection, it is appeared that features['pmid'] and labs['pmid'] are actually equal in values, but not in dtype

Proposed change: convert either of 2 variables to the other's dtype, e.g: in pk_classifier/bootstrap.py, change line 95 from:

assert features['pmid'].equals(labs['pmid'])

to

assert features['pmid'].equals(labs['pmid'].astype('int32'))