logging.info("Beginning summary csv generation")
with open(os.path.join(experiment_directory, "summary.csv"), 'wb') as summary_stream:
writer = csv.writer(summary_stream)
writerow = writer.writerow
format_num = lambda n: n if n is None else "%.3f" % n
if latex_encode:
str_encoder = lambda s: s.encode('latex')
else:
str_encoder = lambda s: s
highlight = lambda x: x
# Get all the union of all the variations names.
variations = uniqueify(itertools.chain(*[r.keys() for r in summary_results.values()]))
data_names = summary_results.keys()
top_results = [max(r.values()) for r in (summary_results[dn] for dn in data_names)]
#None at start to leave column for variation names
writerow([None] + map(str_encoder, data_names))
for variation in variations:
variation_results = [r.get(variation, None) for r in (summary_results[dn] for dn in data_names)]
variation_results_highlighted = [highlight(format_num(v)) if v == t else format_num(v) for (v, t) in zip(variation_results, top_results)]
row = [str_encoder(variation)] + variation_results_highlighted
writerow(row)
logging.info("Ending summary csv generation")
if __name__ == "__main__":
dsn_to_results_dict = OrderedDict()
dataset_names = rows[0][1:]
for dsn in dataset_names:
dsn_to_results_dict[dsn] = []
strats = []
for r in rows[1:]:
strats.append(r[0])
for (v, (dsn, results)) in izip(r[1:], dsn_to_results_dict.iteritems()):
results.append(Result(v, None))
for (dsn, results) in dsn_to_results_dict.iteritems():
result_scores = [r.score for r in results]
result_scores_for_indexing = sorted(uniqueify(result_scores), reverse=True)
for r in results:
score = result_scores_for_indexing.index(r.score) + 1
r.rank = score
with open(output_f, 'wb') as output_s:
writer = csv.writer(output_s)
writer.writerow([""] + dsn_to_results_dict.keys()
+ [options.avg_rank_col_name] if options.add_avg_rank_col else [] )
strats_results = zip(*dsn_to_results_dict.values())
names_to_strats_results = zip(strats, strats_results)
for (strat_name, strat_results) in names_to_strats_results:
if options.include_ranks:
formatter = lambda res: "%.03f (%d)" % (res.score, res.rank)
else:
formatter = lambda res: "%.03f" % res.score
from all_experiments import experiment as all_experiment
from utils import sub_pairs, uniqueify
import baselines, baseline_sparsity, competence_counting, competence_hybrids, competence_with_similarity, competence_with_sparsity
names = (baselines.names + baseline_sparsity.names + competence_counting.names + competence_hybrids.names
+ competence_with_similarity.names + competence_with_sparsity.names)
names = uniqueify(names)
experiment = all_experiment.create_sub_experiment(names)
