def test_sample_model():
"""
Tests sampling function
"""
model = pf.GASLLEV(data=data, family=pf.Normal())
x = model.fit('BBVI', iterations=100)
sample = model.sample(nsims=100)
assert (sample.shape[0] == 100)
assert (sample.shape[1] == len(data) - 1)
def test_poisson_predict_nans():
"""
Tests that the predictions are not nans
"""
model = pf.GASLLEV(data=countdata, family=pf.GASPoisson())
x = model.fit()
x.summary()
assert (len(
model.predict(h=5).values[np.isnan(model.predict(h=5).values)]) == 0)
def test_skewt_couple_terms_integ():
"""
Tests latent variable list length is correct, and that the estimated
latent variables are not nan
"""
model = pf.GASLLEV(data=data, integ=1, family=pf.Skewt())
x = model.fit()
assert (len(model.latent_variables.z_list) == 4)
lvs = np.array([i.value for i in model.latent_variables.z_list])
assert (len(lvs[np.isnan(lvs)]) == 0)
def test_skewt_predict_nans():
"""
Tests that the predictions are not nans
model = pf.GASLLEV(data=data, family=pf.Skewt())
"""
model = pf.GASLLEV(data=data, family=pf.Skewt())
x = model.fit()
x.summary()
assert (len(
model.predict(h=5).values[np.isnan(model.predict(h=5).values)]) == 0)
def test_skewt_pml():
"""
Tests a PML model estimated with Laplace approximation and that the length of the
latent variable list is correct, and that the estimated latent variables are not nan
"""
model = pf.GASLLEV(data=data, family=pf.Skewt())
x = model.fit('PML')
assert (len(model.latent_variables.z_list) == 4)
lvs = np.array([i.value for i in model.latent_variables.z_list])
assert (len(lvs[np.isnan(lvs)]) == 0)
def test_skewt_mh():
"""
Tests an GAS model estimated with Metropolis-Hastings and that the length of the
latent variable list is correct, and that the estimated latent variables are not nan
"""
model = pf.GASLLEV(data=data, family=pf.Skewt())
x = model.fit('M-H', nsims=300)
assert (len(model.latent_variables.z_list) == 4)
lvs = np.array([i.value for i in model.latent_variables.z_list])
assert (len(lvs[np.isnan(lvs)]) == 0)
def test_t_predict_is_nans():
"""
Tests that the in-sample predictions are not nans
"""
model = pf.GASLLEV(data=data, family=pf.t())
x = model.fit()
x.summary()
assert (len(
model.predict_is(h=5).values[np.isnan(
model.predict_is(h=5).values)]) == 0)
def test_poisson_laplace():
"""
Tests an GAS model estimated with Laplace approximation and that the length of the
latent variable list is correct, and that the estimated latent variables are not nan
"""
model = pf.GASLLEV(data=countdata, family=pf.GASPoisson())
x = model.fit('Laplace')
assert (len(model.latent_variables.z_list) == 1)
lvs = np.array([i.value for i in model.latent_variables.z_list])
assert (len(lvs[np.isnan(lvs)]) == 0)
def test_poisson_couple_terms():
"""
Tests latent variable list length is correct, and that the estimated
latent variables are not nan
"""
model = pf.GASLLEV(data=countdata, family=pf.GASPoisson())
x = model.fit()
assert (len(model.latent_variables.z_list) == 1)
lvs = np.array([i.value for i in model.latent_variables.z_list])
assert (len(lvs[np.isnan(lvs)]) == 0)
def test_laplace_bbvi():
"""
Tests an GAS model estimated with BBVI and that the length of the latent variable
list is correct, and that the estimated latent variables are not nan
"""
model = pf.GASLLEV(data=data, family=pf.GASLaplace())
x = model.fit('BBVI', iterations=100)
assert (len(model.latent_variables.z_list) == 2)
lvs = np.array([i.value for i in model.latent_variables.z_list])
assert (len(lvs[np.isnan(lvs)]) == 0)
def test_skewt_bbvi_mini_batch():
"""
Tests an ARIMA model estimated with BBVI and that the length of the latent variable
list is correct, and that the estimated latent variables are not nan
"""
model = pf.GASLLEV(data=data, family=pf.Skewt())
x = model.fit('BBVI', iterations=100, mini_batch=32)
assert (len(model.latent_variables.z_list) == 4)
lvs = np.array([i.value for i in model.latent_variables.z_list])
assert (len(lvs[np.isnan(lvs)]) == 0)
def test_t_bbvi_mini_batch_elbo():
"""
Tests that the ELBO increases
"""
model = pf.GASLLEV(data=data, family=pf.t())
x = model.fit('BBVI',
iterations=100,
mini_batch=32,
record_elbo=True,
map_start=False)
assert (x.elbo_records[-1] > x.elbo_records[0])
def test_t_couple_terms_integ():
"""
Tests an GAS model with 1 AR and 1 MA term, integrated once, and that
the latent variable list length is correct, and that the estimated
latent variables are not nan
"""
model = pf.GASLLEV(data=data, integ=1, family=pf.t())
x = model.fit()
assert (len(model.latent_variables.z_list) == 3)
lvs = np.array([i.value for i in model.latent_variables.z_list])
assert (len(lvs[np.isnan(lvs)]) == 0)
def test_predict_is_intervals_bbvi():
"""
Tests prediction intervals are ordered correctly
"""
model = pf.GASLLEV(data=data, family=pf.Normal())
x = model.fit('BBVI', iterations=100)
predictions = model.predict_is(h=10, intervals=True)
assert (np.all(predictions['99% Prediction Interval'].values >
predictions['95% Prediction Interval'].values))
assert (np.all(predictions['95% Prediction Interval'].values >
predictions['5% Prediction Interval'].values))
assert (np.all(predictions['5% Prediction Interval'].values >
predictions['1% Prediction Interval'].values))
def test_t_predict_is_intervals_mh():
"""
Tests prediction intervals are ordered correctly
"""
model = pf.GASLLEV(data=data, family=pf.t())
x = model.fit('M-H', nsims=400)
predictions = model.predict_is(h=10, intervals=True)
assert (np.all(predictions['99% Prediction Interval'].values >
predictions['95% Prediction Interval'].values))
assert (np.all(predictions['95% Prediction Interval'].values >
predictions['5% Prediction Interval'].values))
assert (np.all(predictions['5% Prediction Interval'].values >
predictions['1% Prediction Interval'].values))
def test_poisson_predict_is_intervals():
"""
Tests prediction intervals are ordered correctly
"""
model = pf.GASLLEV(data=data, family=pf.Poisson())
x = model.fit()
predictions = model.predict_is(h=10, intervals=True)
assert (np.all(predictions['99% Prediction Interval'].values >=
predictions['95% Prediction Interval'].values))
assert (np.all(predictions['95% Prediction Interval'].values >=
predictions['5% Prediction Interval'].values))
assert (np.all(predictions['5% Prediction Interval'].values >=
predictions['1% Prediction Interval'].values))
def build_model(data, ar=2, sc=4, family=pf.families.Poisson, target=None):
model = pf.GASLLEV(data=data, family=family(), target=target)
return model
#print cpu_data
fig2 = plt.figure(figsize=(15, 5))
plt.plot(date, tempf, 'r-', linewidth=2)
plt.xlabel("Timestamp")
plt.ylabel("Filtered CPU Usage (%)")
plt.legend(['Filtered'])
plt.title("Filtered CPU usage")
plt.show()
cpu_data = tempf
print "start ensemble"
model1 = pf.ARIMA(data=cpu_data, ar=4, ma=0)
model2 = pf.ARIMA(data=cpu_data, ar=8, ma=0)
model3 = pf.LLEV(data=cpu_data)
#model4 = pf.GASLLEV(data=cpu_data, family=pf.GASt())
model4 = pf.GASLLEV(data=cpu_data, family=pf.Poisson())
model5 = pf.GPNARX(data=cpu_data, ar=1, kernel=pf.SquaredExponential())
model6 = pf.GPNARX(data=cpu_data, ar=2, kernel=pf.SquaredExponential())
model7 = pf.DynReg('CPUusage', data=dataframe)
mix = pf.Aggregate(learning_rate=1.0, loss_type='squared')
mix.add_model(model1)
mix.add_model(model2)
mix.add_model(model3)
#mix.add_model(model4)
#mix.add_model(model5)
#mix.add_model(model6)
mix.add_model(model7)
mix.tune_learning_rate(20)
print mix.learning_rate
