def run(self):
log.debug("Updating mirror database")
geoip = GeoIP(Config.GEOIP_PATH_V4)
for status in mirror_statuses(
unofficial_mirrors=Config.UNOFFICIAL_MIRRORS):
name = status['mirror']
if name == "a.pypi.python.org":
# don't include 'a' in the list of mirrors - it's no mirror after all
continue
time_diff = status['time_diff']
if not isinstance(time_diff, timedelta):
continue
log.debug(" Processing mirror '%s'", name)
record = geoip.record_by_name(name)
lat = record['latitude']
lon = record['longitude']
log.debug(" Age: %d, Lat: %0.5f, Lon: %0.5f",
time_diff.total_seconds(), lat, lon)
try:
mirror = Mirror.objects.get(name=name)
except ObjectNotFound:
mirror = Mirror(name=name)
mirror.age = time_diff.total_seconds()
mirror.lat = lat
mirror.lon = lon
mirror.save()
self.redis.set(Config.KEY_LAST_UPDATE, time.time())
log.debug("Finished updating mirror database")
def run(self):
log.debug("Updating mirror database")
geoip = GeoIP(Config.GEOIP_PATH_V4)
for status in mirror_statuses(unofficial_mirrors=Config.UNOFFICIAL_MIRRORS):
name = status['mirror']
if name == "a.pypi.python.org":
# don't include 'a' in the list of mirrors - it's no mirror after all
continue
time_diff = status['time_diff']
if not isinstance(time_diff, timedelta):
continue
log.debug(" Processing mirror '%s'", name)
record = geoip.record_by_name(name)
lat = record['latitude']
lon = record['longitude']
log.debug(" Age: %d, Lat: %0.5f, Lon: %0.5f", time_diff.total_seconds(), lat, lon)
try:
mirror = Mirror.objects.get(name=name)
except ObjectNotFound:
mirror = Mirror(name=name)
mirror.age = time_diff.total_seconds()
mirror.lat = lat
mirror.lon = lon
mirror.save()
self.redis.set(Config.KEY_LAST_UPDATE, time.time())
log.debug("Finished updating mirror database")
from scipy.stats import pearsonr
import numpy as np
import transform
import utils
from models import Mirror
from datasets import GlucoseData
from torch.utils.data import DataLoader
cgm_file, meals_file = utils.get_files()
train, val = GlucoseData.train_val_split(cgm_file,
meals_file,
transform=transform.to_tensor)
val_dl = DataLoader(val)
model = Mirror()
amount = 0
acc_pearson = 0.0
for sample in val_dl:
out = model(sample['cgm'], sample['meals'])
out = out.numpy()[0]
target = sample['target'].numpy()[0]
corr, _ = pearsonr(out, target)
if not np.isnan(corr):
amount += 1
acc_pearson += corr
total = acc_pearson / amount
print(f'{total:.2f} mean pearson correlation')
print(f'{len(val) - amount}/{len(val)} were nans')
# 0.06 mean pearson correlation