def compute_conf_scores(path):
logging.info("Computing confidence scores")
job = load_job(path)
raw = job.results.raw_stats
bins = job.results.bins
unperm_counts = cumulative_hist(raw, bins)
perm_counts = job.results.bin_to_mean_perm_count
bin_to_score = confidence_scores(unperm_counts, perm_counts, np.shape(raw)[-1])
feature_to_score = assign_scores_to_features(
raw, bins, bin_to_score)
with h5py.File(path, 'r+') as db:
db.create_dataset("bin_to_unperm_count", data=unperm_counts)
db.create_dataset("bin_to_score", data=bin_to_score)
db.create_dataset("feature_to_score", data=feature_to_score)
def compute_conf_scores(path):
logging.info("Computing confidence scores")
job = load_job(path)
raw = job.results.raw_stats
bins = job.results.bins
unperm_counts = cumulative_hist(raw, bins)
perm_counts = job.results.bin_to_mean_perm_count
bin_to_score = confidence_scores(unperm_counts, perm_counts,
np.shape(raw)[-1])
feature_to_score = assign_scores_to_features(raw, bins, bin_to_score)
with h5py.File(path, 'r+') as db:
db.create_dataset("bin_to_unperm_count", data=unperm_counts)
db.create_dataset("bin_to_score", data=bin_to_score)
db.create_dataset("feature_to_score", data=feature_to_score)
def score_dist_by_tuning_param(job_meta, job_db):
plt.title('Features by confidence score')
plt.xlabel('Confidence')
plt.ylabel('Features')
lines = []
labels = []
params = job_db.settings.tuning_params
if 'tuning_param_idx' in request.args:
params = [int(request.args.get('tuning_param_idx'))]
for i, alpha in enumerate(params):
bins = np.arange(0.5, 1.0, 0.01)
hist = cumulative_hist(job_db.results.feature_to_score[i], bins)
lines.append(plt.plot(bins[:-1], hist, label=str(alpha)))
labels.append(str(alpha))
plt.legend(loc='upper right')
return figure_response()
def score_dist_by_tuning_param(job_meta, job_db):
plt.title('Features by confidence score')
plt.xlabel('Confidence')
plt.ylabel('Features')
lines = []
labels = []
params = job_db.settings.tuning_params
if 'tuning_param_idx' in request.args:
params = [ int(request.args.get('tuning_param_idx')) ]
for i, alpha in enumerate(params):
bins = np.arange(0.5, 1.0, 0.01)
hist = cumulative_hist(job_db.results.feature_to_score[i], bins)
lines.append(plt.plot(bins[:-1], hist, label=str(alpha)))
labels.append(str(alpha))
plt.legend(loc='upper right')
return figure_response()