def test_init(self):
df = pd.DataFrame({
'time': [30, 20, 10],
'item': [3, 2, 1],
'duration': [10, 10, 10]
})
a = AvailabilityFilter(df) #todo: needs experiment environment
r = rs.RecDat()
r.time = 21
def test_init(self):
df = pd.DataFrame({
'time': [30, 20, 10],
'item': [3, 2, 1],
'duration': [10, 10, 10]
})
a = prs.AvailabilityFilter(df)
r = rs.RecDat()
r.time = 21
a.run(r)
def predict(model, user, item):
rd = rs.RecDat()
rd.user = user
rd.item = item
return model.prediction(rd)
import alpenglow.Getter as rs
data = pd.read_csv("../../python/test_alpenglow/test_data_4",
sep=' ',
header=None,
names=['time', 'user', 'item', 'id', 'score', 'eval'])
model = FactorModel(
factor_seed=254938879,
dimension=10,
negative_rate=9,
number_of_iterations=20,
)
model.fit(data)
model.model.write("output_file") #writes model to output_file
rd = rs.RecDat()
rd.user = 3
rd.item = 5
print("prediction for user=3, item=5:", model.model.prediction(rd))
#model2 must have the same dimension
model2 = FactorModel(
factor_seed=1234,
dimension=10,
negative_rate=0,
number_of_iterations=0,
)
#to create the inner model but avoid training, we need to run fit()
#on an empty dataset
data2 = pd.DataFrame(columns=['time', 'user', 'item'])
model2.fit(data2)
