def handle_stats( stats_counter ):
stats_counter_thresh=40
# Updating stats for rpisurv community every 40 loops
if stats_counter % stats_counter_thresh == 0:
stats.update_stats(uniqid, str(stats.get_runtime(start_time)), update_stats_enabled)
else:
logger.debug("stats_counter is " + str(stats_counter) + ". Only sending every " + str(stats_counter_thresh))
def handle_stats( stats_counter ):
stats_counter_thresh=40
# Updating stats for rpisurv community every 40 loops
if stats_counter % stats_counter_thresh == 0:
stats.update_stats(uniqid, str(stats.get_runtime(start_time)), update_stats_enabled)
else:
logger.debug("stats_counter is " + str(stats_counter) + ". Only sending every " + str(stats_counter_thresh))
def update():
more = True
while more:
stat_session = db.StatSession()
log_session = db.LogSession()
events, more = logparser.parse_sql(log_session)
stats.update_stats(stat_session, events)
stat_session.commit()
log_session.commit()
def hidden_stats(ticker, adj, mar, n):
## Checking delist.csv
file = os.getcwd() + '\\data\\delist.csv'
dlf = pd.read_csv(file)
delist_date = dlf.columns[0]
td = (dateparser.parse(str(today)) - dateparser.parse(delist_date))
# If delist.csv has not been updated since yesterday, then update.
if td > timedelta(days=1):
print('Retrieving a new list of delisted stocks...')
delist.update_delist()
# Adjusting values based on slider from the top dashboard
st = stats.update_stats(ticker)
if st != None and st != 'NoneType':
if adj != 0 or mar != 5:
st['parsed']['Entry Point (Decimal)'] = st['parsed'][
'Entry Point (Decimal)'] + (adj / 100)
st['parsed']['Entry Point'] = '{}%'.format(
st['parsed']['Entry Point (Decimal)'] * 100)
st['notes']['Engage?'] = '{} ({})'.format(
st['parsed']['Engage?'], st['parsed']['Entry Point'])
st = json.dumps(st, indent=4)
return (st)
def test_model(trained_model, test_data, language_stats, device,
language_names_dict, int2lang):
correct_per_example = 0
total_predictions = 5000
incorrect_guesses_per_instance = 0
percent = 0
example = 0
batch_nr = 1
tenp = 500
num_characters = []
count = 0
for x, y in tqdm(test_data):
batch_nr += 1
example += 1
hidden_layer = trained_model.init_hidden(1).to(device)
for examples in zip(x, y):
#total_predictions += 1
count += 1
prediction = trained_model(examples[0].unsqueeze(0).to(device),
hidden_layer)
_, indeces = torch.max(prediction[0].data, dim=1)
characters = len(torch.nonzero(examples[0]))
if indeces[0].item() == examples[1].item():
num_characters.append(characters)
correct_per_example += 1
stats.update_stats(language_stats, indeces[0].item(),
examples[1].item(), int2lang, characters,
language_names_dict)
break
else:
#characters = 0
stats.update_stats(language_stats, indeces[0].item(),
examples[1].item(), int2lang, characters,
language_names_dict)
incorrect_guesses_per_instance += 1
continue
print(example)
print(correct_per_example)
print(incorrect_guesses_per_instance)
print(count)
return language_stats
def compute_update_stats(writer, step, modules, prev_params):
for name, module in modules.items():
update_scales, param_scales, fractions = stats.update_stats(
get_cpu_params(module), prev_params[name])
writer.add_scalar(
f'update/{name}_mean_update_scale', np.mean(update_scales), step)
writer.add_scalar(
f'update/{name}_mean_param_scale', np.mean(param_scales), step)
writer.add_scalar(
f'update/{name}_mean_update_to_param_scale', np.mean(fractions), step)
writer.add_histogram(
f'update/{name}_update_to_param_scale', fractions, step)
writer.add_histogram(
f'update/{name}_update_scale', update_scales, step)
writer.add_histogram(
f'update/{name}_param_scale', param_scales, step)
def create_new_descriptive_stats():
data = convert_to_string(request.data)
descriptive, correlations = stats.update_stats(data)
cache.set("descriptive", descriptive)
cache.set("correlations", correlations)
return "Successfully updated statistics cache", 200