import datetime
all_check_ins = train + test
all_venues = [x["venue_id"] for x in all_check_ins]
n_values = Counter(all_venues)
coordinates = {}
for venue in all_venues:
latitude = np.median([
x["latitude"] for x in all_check_ins
if x["venue_id"] == venue
])
longitude = np.median([
x["longitude"] for x in all_check_ins
if x["venue_id"] == venue
])
coordinates[venue] = (latitude, longitude)
model = NCGModel(n_values, coordinates)
model.train(train, number_of_iterations=10)
if model.parameters != None:
correct = 0
for check_in in test:
real_venue = check_in["venue_id"]
time = check_in["date"]
predicted_venue = model.predict(time, train + test)
if real_venue == predicted_venue:
correct += 1
results_radiation.append(float(correct) / len(test))
if user in network:
social_model_stanford.fit_model(model, network[user],
datasets)
correct = 0
