def impute_gain(x, odir):
python_exe = 'python3'
script = '{}/gain/gain.py'.format(utilmlab.get_proj_dir())
fn_i = '{}/xmissing.csv.gz'.format(odir)
fn_o = '{}/ximputed.csv'.format(odir)
x.to_csv(fn_i, compression='gzip', index=False)
if os.path.isfile(fn_o):
os.remove(fn_o)
utilmlab.exe_cmd(
logger, '{} {} -i {} -o {} --testall 1'.format(python_exe, script,
fn_i, fn_o))
return pd.read_csv(fn_o)
numexp = args.numexp
proj_dir = utilmlab.get_proj_dir() \
if args.projdir is None else args.projdir
alg = 'asac'
if args.exe is not None:
python_exe = args.exe
else:
python_exe = 'python' if sys.version_info[0] < 3 else 'python3'
niter = args.it
version = 1
resdir = '{}/result/{}/v_{}/h_{}'.format(
proj_dir, alg, version,
os.environ['HOSTNAME'] if 'HOSTNAME' in os.environ else 'unknown')
utilmlab.ensure_dir(resdir)
logger = utilmlab.init_logger(resdir, 'log_test_{}.txt'.format(alg))
result_lst = []
script = Path('{}/alg/asac/Main_Synthetic_Exp1.py'.format(proj_dir))
utilmlab.exe_cmd(
logger, '{} {} {}'.format(python_exe, script,
'--it {} -n {}'.format(niter, numexp)))
if dataset == 'bc':
# from sklearn: A copy of UCI ML Breast
# Cancer Wisconsin (Diagnostic) dataset
x, y = load_breast_cancer(return_X_y=True)
elif dataset == 'cover':
x, y = fetch_covtype(return_X_y=True)
else:
assert 0
lbl = 'target'
df = pd.DataFrame(x)
df[lbl] = y
df.to_csv(fn_csv, index=False, compression='gzip', sep=sep)
utilmlab.exe_cmd(
logger,
'{} {} -i {} --target {} -o {} --verbose {} --it {} -n {} --separator {}'
.format(python_exe, script, fn_csv, lbl, resdir, verbose, niter,
nsample, sep))
utilmlab.exe_cmd(
logger,
'{} {} -i {} -o {} --verbose {} --it {} --target {} --model {} --separator {}'
' --nstage {} -n {}'.format(
python_exe,
script,
fn_csv,
resdir,
verbose,
niter,
lbl,
fn_model,
dataset = 'bc'
odir = '{}/misc/dataset_{}'.format(resdir, dataset)
fn_csv, fn_missing_csv, fn_imputed_csv = set_filenames(odir)
script_create_missing = Path(
'{}/alg/gain/create_missing.py'.format(proj_dir))
script = Path('{}/alg/gain/gain.py'.format(proj_dir))
script_ana = Path('{}/alg/gain/gain_ana.py'.format(proj_dir))
for islabel in [0, 1]:
for autocat in [0, 1, 2]:
utilmlab.exe_cmd(
logger, '{} {} --dataset {} -o {} '
'--oref {} --istarget {}'.format(python_exe,
script_create_missing,
dataset, fn_missing_csv,
fn_csv, islabel))
utilmlab.exe_cmd(
logger, '{} {} -i {} {} '
'-o {} --it {} --testall 1 --autocategorical {}'.format(
python_exe, script, fn_missing_csv,
'--target target' if islabel else '', fn_imputed_csv,
niter, autocat))
result_lst = []
dataset_prop = [('spambase', None), ('spambase', 'label'), ('bc', None),
('spam', None), ('letter-recognition', None),
('letter-recognition', 'lettr'), ('letter', None)]
if not os.path.isfile('{}/alg/gcit/ccle_experiments/data/mutation.txt.gz'.
format(proj_dir)):
print('warning: data files for ccle_experiments not found')
sys.exit(0)
alg = 'gcit'
version = 1
if args.exe is not None:
python_exe = args.exe
else:
python_exe = 'python' if sys.version_info[0] < 3 else 'python3'
if args.o is None:
resdir = '{}/result/{}/v_{}/h_{}'.format(
proj_dir, alg, version,
os.environ['HOSTNAME'] if 'HOSTNAME' in os.environ else 'unknown')
else:
resdir = args.o
utilmlab.ensure_dir(resdir)
logger = utilmlab.init_logger(resdir, 'log_test_{}.txt'.format(alg))
script = Path('{}/alg/gcit/ccle_experiments/'
'ccle_experiment.py'.format(proj_dir))
odir = resdir
utilmlab.exe_cmd(logger, '{} {}'.format(python_exe, script))
# execute all unit tests
f_lst = utilmlab.find_file_dir(
'{}/alg'.format(proj_dir),
'test_*.py')
logger.info('Unit tests found:{}'.format(f_lst))
for fpy in f_lst:
if 'test_alg.py' in fpy:
continue
utilmlab.exe_cmd(
logger,
'{} {} {} {}'.format(
python_exe,
Path(fpy),
'--it {}'.format(args.it) if args.it is not None else '',
'--exe {}'.format(args.exe) if args.exe is not None else ''
)
)
# execute all notebooks
f_lst = utilmlab.find_file_dir(
'{}/alg'.format(utilmlab.get_proj_dir()),
'*.ipynb')
logger.info('notebooks found:{}'.format(f_lst))
cwd = os.getcwd()
if not is_only_notebook:
# execute all unit tests
f_lst = utilmlab.find_file_dir(test_dir, 'test_*.py')
logger.info('Unit tests found:{}'.format(f_lst))
for fpy in f_lst:
if 'test_alg.py' in fpy:
continue
time_start = time.time()
utilmlab.exe_cmd(
logger, '{} {} {} {}'.format(
python_exe, Path(fpy),
'--it {}'.format(args.it) if args.it is not None else '',
'--exe {}'.format(args.exe)
if args.exe is not None else ''))
logger.info('time={}'.format(time.time() - time_start))
# execute all notebooks
f_lst = utilmlab.find_file_dir(test_dir, '*.ipynb')
logger.info('notebooks found:{}'.format(f_lst))
cwd = os.getcwd()
fn_html_lst = []
for fnb in f_lst:
elif dataset == 'cover':
x, y = fetch_covtype(return_X_y=True)
else:
assert 0
lbl = 'target'
df = pd.DataFrame(x)
df[lbl] = y
df.to_csv(fn_csv, index=False, compression='gzip', sep=sep)
try:
utilmlab.exe_cmd(
logger,
'Rscript {} -i {} --target {} --exe {} --it {} '
' --replication {} --projdir {}'.format(
script, fn_csv, lbl, python_exe, niter, nreplication,
proj_dir),
assert_on_error=
not generate_error # assert if an error is not expected
)
except:
if generate_error:
logger.info('expected error generated')
pass
assert 0
fn_data_csv = '{}/data.csv'.format(resdir)
fn_json = '{}/generated_data_properties.json'.format(resdir)
utilmlab.exe_cmd(
logger, 'Rscript {}/alg/knockoffgan/gen_data.r -o {} --target {} '
for dataset in ['csv', 'bc', 'spambase']:
odir = '{}/dataset_{}'.format(resdir, dataset)
utilmlab.ensure_dir(odir)
fn_feature_score = '{}/feature_score.csv.gz'.format(odir)
fn_json = '{}/feature_score.csv.json'.format(odir)
fn_plot_sample = '{}/sample.png'.format(odir)
fn_plot_global = '{}/global.png'.format(odir)
if dataset == 'csv':
fn_csv = '{}/spambase.csv.gz'.format(utilmlab.get_data_dir())
utilmlab.exe_cmd(
logger, '{} {} -i {} --target label --it {} -o {}'.format(
python_exe, script, fn_csv, nepoch, fn_feature_score))
else:
if not data_loader_mlab.is_available(dataset):
continue
utilmlab.exe_cmd(
logger, '{} {} --dataset {} --it {} -o {}'.format(
python_exe, script, dataset, nepoch, fn_feature_score))
utilmlab.exe_cmd(
logger, '{} {} -i {} -o {}'.format(python_exe, script_ana,
fn_feature_score, fn_json))
utilmlab.exe_cmd(
logger, '{} {} -i {} -oglobal {} -osample {}'.format(
python_exe, script_plot, fn_feature_score, fn_plot_global,
'--{} {}'.format(el, args_d[el]) for el in args_d.keys()])
dataset = args_d['dataset']
odir = '{}/dataset_{}'.format(
odir,
dataset)
utilmlab.ensure_dir(odir)
utilmlab.exe_cmd(
logger,
'{} {} {} {} {} {} {}'.format(
python_exe,
script,
cmd_arg,
'--it {}'.format(niter),
'-o {}'.format(odir),
'--itout {}'.format(
args.itout) if args.itout is not None else '',
'--itrs {}'.format(
args.itrs) if args.itrs is not None else ''))
utilmlab.exe_cmd(
logger,
'{} {} {} {} {}'.format(
python_exe,
script_ana,
cmd_arg,
'-o {}'.format(odir),
'--itout {}'.format(
args.itout) if args.itout is not None else ''))