def link_incompatible_error(ip, port):
print("Reading in original prostate data.")
prostate = h2o.import_file(
path=h2o.locate("smalldata/prostate/prostate.csv.zip"))
print(
"Throw error when trying to create model with incompatible logit link."
)
try:
h2o.model = h2o.glm(x=prostate[1:8],
y=prostate[8],
family="gaussian",
link="logit")
assert False, "expected an error"
except EnvironmentError:
assert True
try:
h2o.model = h2o.glm(x=prostate[1:8],
y=prostate[8],
family="tweedie",
link="log")
assert False, "expected an error"
except EnvironmentError:
assert True
try:
h2o.model = h2o.glm(x=prostate[2:9],
y=prostate[1],
family="binomial",
link="inverse")
assert False, "expected an error"
except EnvironmentError:
assert True
def link_incompatible_error():
print("Reading in original prostate data.")
prostate = h2o.import_file(path=pyunit_utils.locate("smalldata/prostate/prostate.csv.zip"))
print("Throw error when trying to create model with incompatible logit link.")
try:
h2o.model = h2o.glm(x=prostate[1:8], y=prostate[8], family="gaussian", link="logit")
assert False, "expected an error"
except EnvironmentError:
assert True
try:
h2o.model = h2o.glm(x=prostate[1:8], y=prostate[8], family="tweedie", link="log")
assert False, "expected an error"
except EnvironmentError:
assert True
try:
h2o.model = h2o.glm(x=prostate[2:9], y=prostate[1], family="binomial", link="inverse")
assert False, "expected an error"
except EnvironmentError:
assert True
def covtype_get_model(ip,port):
# Connect to h2o
h2o.init(ip,port)
#Log.info("Importing covtype.20k.data...\n")
covtype = h2o.import_frame(path=h2o.locate("smalldata/covtype/covtype.20k.data"))
Y = 54
X = range(0,20) + range(29,54)
# Set response to be indicator of a particular class
res_class = random.randint(1,4)
# Log.info(paste("Setting response column", myY, "to be indicator of class", res_class, "\n"))
covtype[54] = (covtype[54] == res_class)
#covtype_data.summary()
# L2: alpha = 0, lambda = 0
covtype_mod1 = h2o.glm(y=covtype[Y], x=covtype[X], family="binomial", alpha=[0], Lambda=[0])
covtype_mod1.show()
covtype_mod1 = h2o.get_model(covtype_mod1._id)
covtype_mod1.show()
# Elastic: alpha = 0.5, lambda = 1e-4
covtype_mod2 = h2o.glm(y=covtype[Y], x=covtype[X], family="binomial", alpha=[0.5], Lambda=[1e-4])
covtype_mod2.show()
covtype_mod2 = h2o.get_model(covtype_mod2._id)
covtype_mod2.show()
# L1: alpha = 1, lambda = 1e-4
covtype_mod3 = h2o.glm(y=covtype[Y], x=covtype[X], family="binomial", alpha=[1], Lambda=[1e-4])
covtype_mod3.show()
covtype_mod3 = h2o.get_model(covtype_mod3._id)
covtype_mod3.show()
def getLambdaModel():
print("Read data")
prostate = h2o.import_file(path=tests.locate("smalldata/logreg/prostate.csv"))
myX = ["AGE","RACE","DPROS","DCAPS","PSA","VOL","GLEASON"]
myY = "CAPSULE"
family = random.choice(["gaussian","binomial"])
print(family)
print("Do lambda search and build models")
if family == "gaussian":
model = h2o.glm(x=prostate[myX], y=prostate[myY], family=family, standardize=True, use_all_factor_levels=True, lambda_search=True)
else:
model = h2o.glm(x=prostate[myX], y=prostate[myY].asfactor(), family=family, standardize=True, use_all_factor_levels=True, lambda_search=True)
print("the models were built over the following lambda values: ")
all_lambdas = model.models(1).lambda_all()
print(all_lambdas)
for i in range(10):
Lambda = random.sample(all_lambdas,1)
print("For Lambda we get this model:")
m1 = h2o.getGLMLambdaModel(model.models(random.randint(0,len(model.models()-1)),Lambda=Lambda))
m1.show()
print("this model should be same as the one above:")
m2 = h2o.getGLMLambdaModel(model.models(random.randint(0,len(model.models()-1)),Lambda=Lambda))
m2.show()
assert m1==m2, "expected models to be equal"
def link_functions_gamma(ip, port):
# Connect to h2o
h2o.init(ip, port)
print("Read in prostate data.")
h2o_data = h2o.import_frame(
path=h2o.locate("smalldata/prostate/prostate_complete.csv.zip"))
h2o_data.head()
sm_data = pd.read_csv(
zipfile.ZipFile(
h2o.locate("smalldata/prostate/prostate_complete.csv.zip")).open(
"prostate_complete.csv")).as_matrix()
sm_data_response = sm_data[:, 5]
sm_data_features = sm_data[:, [1, 2, 3, 4, 6, 7, 8, 9]]
print("Testing for family: GAMMA")
print("Set variables for h2o.")
myY = "DPROS"
myX = ["ID", "AGE", "RACE", "GLEASON", "DCAPS", "PSA", "VOL", "CAPSULE"]
print("Create models with canonical link: INVERSE")
h2o_model_in = h2o.glm(x=h2o_data[myX],
y=h2o_data[myY],
family="gamma",
link="inverse",
alpha=[0.5],
Lambda=[0],
n_folds=0)
sm_model_in = sm.GLM(endog=sm_data_response,
exog=sm_data_features,
family=sm.families.Gamma(
sm.families.links.inverse_power)).fit()
print("Compare model deviances for link function inverse")
h2o_deviance_in = h2o_model_in._model_json['output'][
'residual_deviance'] / h2o_model_in._model_json['output'][
'null_deviance']
sm_deviance_in = sm_model_in.deviance / sm_model_in.null_deviance
assert h2o_deviance_in - sm_deviance_in < 0.01, "expected h2o to have an equivalent or better deviance measures"
print("Create models with canonical link: LOG")
h2o_model_log = h2o.glm(x=h2o_data[myX],
y=h2o_data[myY],
family="gamma",
link="log",
alpha=[0.5],
Lambda=[0],
n_folds=0)
sm_model_log = sm.GLM(endog=sm_data_response,
exog=sm_data_features,
family=sm.families.Gamma(
sm.families.links.log)).fit()
print("Compare model deviances for link function log")
h2o_deviance_log = h2o_model_log._model_json['output'][
'residual_deviance'] / h2o_model_log._model_json['output'][
'null_deviance']
sm_deviance_log = sm_model_log.deviance / sm_model_log.null_deviance
assert h2o_deviance_log - sm_deviance_log < 0.01, "expected h2o to have an equivalent or better deviance measures"
def link_functions_gamma(ip,port):
# Connect to h2o
h2o.init(ip,port)
print("Read in prostate data.")
h2o_data = h2o.import_frame(path=h2o.locate("smalldata/prostate/prostate_complete.csv.zip"))
h2o_data.head()
sm_data = pd.read_csv(zipfile.ZipFile(h2o.locate("smalldata/prostate/prostate_complete.csv.zip")).open("prostate_complete.csv")).as_matrix()
sm_data_response = sm_data[:,5]
sm_data_features = sm_data[:,[1,2,3,4,6,7,8,9]]
print("Testing for family: GAMMA")
print("Set variables for h2o.")
myY = "DPROS"
myX = ["ID","AGE","RACE","GLEASON","DCAPS","PSA","VOL","CAPSULE"]
print("Create models with canonical link: INVERSE")
h2o_model_in = h2o.glm(x=h2o_data[myX], y=h2o_data[myY], family="gamma", link="inverse",alpha=[0.5], Lambda=[0])
sm_model_in = sm.GLM(endog=sm_data_response, exog=sm_data_features, family=sm.families.Gamma(sm.families.links.inverse_power)).fit()
print("Compare model deviances for link function inverse")
h2o_deviance_in = h2o_model_in._model_json['output']['residual_deviance'] / h2o_model_in._model_json['output']['null_deviance']
sm_deviance_in = sm_model_in.deviance / sm_model_in.null_deviance
assert h2o_deviance_in - sm_deviance_in < 0.01, "expected h2o to have an equivalent or better deviance measures"
print("Create models with canonical link: LOG")
h2o_model_log = h2o.glm(x=h2o_data[myX], y=h2o_data[myY], family="gamma", link="log",alpha=[0.5], Lambda=[0])
sm_model_log = sm.GLM(endog=sm_data_response, exog=sm_data_features, family=sm.families.Gamma(sm.families.links.log)).fit()
print("Compare model deviances for link function log")
h2o_deviance_log = h2o_model_log._model_json['output']['residual_deviance'] / h2o_model_log._model_json['output']['null_deviance']
sm_deviance_log = sm_model_log.deviance / sm_model_log.null_deviance
assert h2o_deviance_log - sm_deviance_log < 0.01, "expected h2o to have an equivalent or better deviance measures"
def shuffling_large(ip,port):
# Connect to h2o
h2o.init(ip,port)
print("Reading in Arcene training data for binomial modeling.")
train_data = h2o.upload_file(path=h2o.locate("smalldata/arcene/shuffle_test_version/arcene.csv"))
train_data_shuffled = h2o.upload_file(path=h2o.locate("smalldata/arcene/shuffle_test_version/arcene_shuffled.csv"))
print("Create model on original Arcene dataset.")
h2o_model = h2o.glm(x=train_data[0:1000], y=train_data[1000], family="binomial", lambda_search=True, alpha=[0.5], n_folds=0, use_all_factor_levels=True)
print("Create second model on original Arcene dataset.")
h2o_model_2 = h2o.glm(x=train_data[0:1000], y=train_data[1000], family="binomial", lambda_search=True, alpha=[0.5], n_folds=0, use_all_factor_levels=True)
print("Create model on shuffled Arcene dataset.")
h2o_model_s = h2o.glm(x=train_data_shuffled[0:1000], y=train_data_shuffled[1000], family="binomial", lambda_search=True, alpha=[0.5], n_folds=0, use_all_factor_levels=True)
print("Assert that number of predictors remaining and their respective coefficients are equal.")
for x, y in zip(h2o_model._model_json['output']['coefficients_table'].cell_values,h2o_model_2._model_json['output']['coefficients_table'].cell_values):
assert (type(x[1]) == type(y[1])) and (type(x[2]) == type(y[2])), "coefficients should be the same type"
if isinstance(x[1],float):
assert abs(x[1] - y[1]) < 5e-10, "coefficients should be equal"
if isinstance(x[2],float):
assert abs(x[2] - y[2]) < 5e-10, "coefficients should be equal"
for x, y in zip(h2o_model._model_json['output']['coefficients_table'].cell_values,h2o_model_s._model_json['output']['coefficients_table'].cell_values):
assert (type(x[1]) == type(y[1])) and (type(x[2]) == type(y[2])), "coefficients should be the same type"
if isinstance(x[1],float):
assert abs(x[1] - y[1]) < 5e-10, "coefficients should be equal"
if isinstance(x[2],float):
assert abs(x[2] - y[2]) < 5e-10, "coefficients should be equal"
def covtype(ip,port):
# Connect to h2o
h2o.init(ip,port)
# Log.info("Importing covtype.20k.data...\n")
covtype = h2o.import_frame(path=h2o.locate("smalldata/covtype/covtype.20k.data"))
#
myY = 54
myX = [x for x in range(0,54) if x not in [20,28]]
# Set response to be indicator of a particular class
res_class = random.randint(1,4)
# Log.info(paste("Setting response column", myY, "to be indicator of class", res_class, "\n"))
covtype[54] = (covtype[54] == res_class)
#covtype.summary()
# L2: alpha = 0, lambda = 0
covtype_mod1 = h2o.glm(y=covtype[myY], x=covtype[myX], family="binomial", n_folds=0, alpha=[0], Lambda=[0])
covtype_mod1.show()
# Elastic: alpha = 0.5, lambda = 1e-4
covtype_mod2 = h2o.glm(y=covtype[myY], x=covtype[myX], family="binomial", n_folds=0, alpha=[0.5], Lambda=[1e-4])
covtype_mod2.show()
# L1: alpha = 1, lambda = 1e-4
covtype_mod3 = h2o.glm(y=covtype[myY], x=covtype[myX], family="binomial", n_folds=0, alpha=[1], Lambda=[1e-4])
covtype_mod3.show()
def getLambdaModel(ip,port):
print("Read data")
prostate = h2o.import_file(path=h2o.locate("smalldata/logreg/prostate.csv"))
myX = ["AGE","RACE","DPROS","DCAPS","PSA","VOL","GLEASON"]
myY = "CAPSULE"
family = random.choice(["gaussian","binomial"])
print(family)
print("Do lambda search and build models")
if family == "gaussian":
model = h2o.glm(x=prostate[myX], y=prostate[myY], family=family, standardize=True, use_all_factor_levels=True, lambda_search=True)
else:
model = h2o.glm(x=prostate[myX], y=prostate[myY].asfactor(), family=family, standardize=True, use_all_factor_levels=True, lambda_search=True)
print("the models were built over the following lambda values: ")
all_lambdas = model.models(1).lambda_all()
print(all_lambdas)
for i in range(10):
Lambda = random.sample(all_lambdas,1)
print("For Lambda we get this model:")
m1 = h2o.getGLMLambdaModel(model.models(random.randint(0,len(model.models()-1)),Lambda=Lambda))
m1.show()
print("this model should be same as the one above:")
m2 = h2o.getGLMLambdaModel(model.models(random.randint(0,len(model.models()-1)),Lambda=Lambda))
m2.show()
assert m1==m2, "expected models to be equal"
def lambda_search(ip,port):
# Connect to h2o
h2o.init(ip,port)
#Log.info("Importing prostate.csv data...\n")
prostate = h2o.import_frame(h2o.locate("smalldata/logreg/prostate.csv"))
#prostate.summary()
# GLM without lambda search, lambda is single user-provided value
#Log.info("H2O GLM (binomial) with parameters: lambda_search = TRUE, nfolds: 2\n")
prostate_nosearch = h2o.glm(x=prostate[2:9], y=prostate[1], training_frame = prostate.hex, family = "binomial", nlambdas = 5, lambda_search = False, n_folds = 2)
params_nosearch = prostate_nosearch.params()
try:
prostate_nosearch.getGLMLambdaModel(0.5)
assert False, "expected an error"
except EnvironmentError:
assert True
# GLM with lambda search, return only model corresponding to best lambda as determined by H2O
#Log.info("H2O GLM (binomial) with parameters: lambda_search: TRUE, nfolds: 2\n")
prostate_search = h2o.glm(x=prostate[2:9], y=prostate[1], training_frame = prostate.hex, family = "binomial", nlambdas = 5, lambda_search = True, n_folds = 2)
params_search = prostate_search.params()
random_lambda = random.choice(prostate_search.lambda_all())
#Log.info(cat("Retrieving model corresponding to randomly chosen lambda", random_lambda, "\n"))
random_model = prostate_search.getGLMLambdaModel(random_lambda)
assert random_model.getLambda() == random_lambda, "expected equal lambdas"
#Log.info(cat("Retrieving model corresponding to best lambda", params.bestlambda$lambda_best, "\n"))
best_model = prostate_search.getGLMLambdaModel(params_search.bestlambda())
assert best_model.model() == prostate_search.model(), "expected models to be equal"
def covtype(ip,port):
# Log.info("Importing covtype.20k.data...\n")
covtype = h2o.import_file(path=h2o.locate("smalldata/covtype/covtype.20k.data"))
#
myY = 54
myX = [x for x in range(0,54) if x not in [20,28]]
# Set response to be indicator of a particular class
res_class = random.randint(1,4)
# Log.info(paste("Setting response column", myY, "to be indicator of class", res_class, "\n"))
covtype[54] = (covtype[54] == res_class)
#covtype.summary()
# L2: alpha = 0, lambda = 0
covtype_mod1 = h2o.glm(y=covtype[myY], x=covtype[myX], family="binomial", alpha=[0], Lambda=[0])
covtype_mod1.show()
# Elastic: alpha = 0.5, lambda = 1e-4
covtype_mod2 = h2o.glm(y=covtype[myY], x=covtype[myX], family="binomial", alpha=[0.5], Lambda=[1e-4])
covtype_mod2.show()
# L1: alpha = 1, lambda = 1e-4
covtype_mod3 = h2o.glm(y=covtype[myY], x=covtype[myX], family="binomial", alpha=[1], Lambda=[1e-4])
covtype_mod3.show()
def shuffling_large(ip,port):
# Connect to h2o
h2o.init(ip,port)
print("Reading in Arcene training data for binomial modeling.")
train_data = h2o.upload_file(path=h2o.locate("smalldata/arcene/shuffle_test_version/arcene.csv"))
train_data_shuffled = h2o.upload_file(path=h2o.locate("smalldata/arcene/shuffle_test_version/arcene_shuffled.csv"))
print("Create model on original Arcene dataset.")
h2o_model = h2o.glm(x=train_data[0:1000], y=train_data[1000], family="binomial", lambda_search=True, alpha=[0.5], use_all_factor_levels=True)
print("Create second model on original Arcene dataset.")
h2o_model_2 = h2o.glm(x=train_data[0:1000], y=train_data[1000], family="binomial", lambda_search=True, alpha=[0.5], use_all_factor_levels=True)
print("Create model on shuffled Arcene dataset.")
h2o_model_s = h2o.glm(x=train_data_shuffled[0:1000], y=train_data_shuffled[1000], family="binomial", lambda_search=True, alpha=[0.5], use_all_factor_levels=True)
print("Assert that number of predictors remaining and their respective coefficients are equal.")
for x, y in zip(h2o_model._model_json['output']['coefficients_table'].cell_values,h2o_model_2._model_json['output']['coefficients_table'].cell_values):
assert (type(x[1]) == type(y[1])) and (type(x[2]) == type(y[2])), "coefficients should be the same type"
if isinstance(x[1],float):
assert abs(x[1] - y[1]) < 5e-10, "coefficients should be equal"
if isinstance(x[2],float):
assert abs(x[2] - y[2]) < 5e-10, "coefficients should be equal"
for x, y in zip(h2o_model._model_json['output']['coefficients_table'].cell_values,h2o_model_s._model_json['output']['coefficients_table'].cell_values):
assert (type(x[1]) == type(y[1])) and (type(x[2]) == type(y[2])), "coefficients should be the same type"
if isinstance(x[1],float):
assert abs(x[1] - y[1]) < 5e-10, "coefficients should be equal"
if isinstance(x[2],float):
assert abs(x[2] - y[2]) < 5e-10, "coefficients should be equal"
def covtype_get_model():
#Log.info("Importing covtype.20k.data...\n")
covtype = h2o.import_file(path=pyunit_utils.locate("smalldata/covtype/covtype.20k.data"))
Y = 54
X = list(range(0,20)) + list(range(29,54))
# Set response to be indicator of a particular class
res_class = random.randint(1,4)
# Log.info(paste("Setting response column", myY, "to be indicator of class", res_class, "\n"))
covtype[54] = (covtype[54] == res_class)
#covtype_data.summary()
# L2: alpha = 0, lambda = 0
covtype_mod1 = h2o.glm(y=covtype[Y], x=covtype[X], family="binomial", alpha=[0], Lambda=[0])
covtype_mod1.show()
covtype_mod1 = h2o.get_model(covtype_mod1._id)
covtype_mod1.show()
# Elastic: alpha = 0.5, lambda = 1e-4
covtype_mod2 = h2o.glm(y=covtype[Y], x=covtype[X], family="binomial", alpha=[0.5], Lambda=[1e-4])
covtype_mod2.show()
covtype_mod2 = h2o.get_model(covtype_mod2._id)
covtype_mod2.show()
# L1: alpha = 1, lambda = 1e-4
covtype_mod3 = h2o.glm(y=covtype[Y], x=covtype[X], family="binomial", alpha=[1], Lambda=[1e-4])
covtype_mod3.show()
covtype_mod3 = h2o.get_model(covtype_mod3._id)
covtype_mod3.show()
def link_functions_poisson():
print("Read in prostate data.")
h2o_data = h2o.import_file(path=pyunit_utils.locate(
"smalldata/prostate/prostate_complete.csv.zip"))
sm_data = pd.read_csv(
zipfile.ZipFile(
pyunit_utils.locate("smalldata/prostate/prostate_complete.csv.zip")
).open("prostate_complete.csv")).as_matrix()
sm_data_response = sm_data[:, 9]
sm_data_features = sm_data[:, 1:9]
print("Testing for family: POISSON")
print("Set variables for h2o.")
myY = "GLEASON"
myX = ["ID", "AGE", "RACE", "CAPSULE", "DCAPS", "PSA", "VOL", "DPROS"]
print("Create h2o model with canonical link: LOG")
h2o_model_log = h2o.glm(x=h2o_data[myX],
y=h2o_data[myY],
family="poisson",
link="log",
alpha=[0.5],
Lambda=[0])
print("Create statsmodel model with canonical link: LOG")
sm_model_log = sm.GLM(endog=sm_data_response,
exog=sm_data_features,
family=sm.families.Poisson(
sm.families.links.log)).fit()
print("Compare model deviances for link function log")
h2o_deviance_log = old_div(h2o_model_log.residual_deviance(),
h2o_model_log.null_deviance())
sm_deviance_log = old_div(sm_model_log.deviance,
sm_model_log.null_deviance)
assert h2o_deviance_log - sm_deviance_log < 0.01, "expected h2o to have an equivalent or better deviance measures"
print("Create h2o models with link: IDENTITY")
h2o_model_id = h2o.glm(x=h2o_data[myX],
y=h2o_data[myY],
family="poisson",
link="identity",
alpha=[0.5],
Lambda=[0])
print("Create statsmodel models with link: IDENTITY")
sm_model_id = sm.GLM(endog=sm_data_response,
exog=sm_data_features,
family=sm.families.Poisson(
sm.families.links.identity)).fit()
print("Compare model deviances for link function identity")
h2o_deviance_id = old_div(h2o_model_id.residual_deviance(),
h2o_model_id.null_deviance())
sm_deviance_id = old_div(sm_model_id.deviance, sm_model_id.null_deviance)
assert h2o_deviance_id - sm_deviance_id < 0.01, "expected h2o to have an equivalent or better deviance measures"
def attack(family, train, valid, x, y):
kwargs = {}
kwargs['family'] = family
gaussian_links = ["inverse", "log", "identity"]
binomial_links = ["logit"]
poisson_links = ["log", "identity"]
gamma_links = ["inverse", "log", "identity"]
# randomly select parameters and their corresponding values
if random.randint(0,1): kwargs['max_iterations'] = random.randint(1,50)
if random.random() > 0.8: kwargs['beta_epsilon'] = random.random()
if random.randint(0,1): kwargs['solver'] = ["AUTO", "IRLSM", "L_BFGS", "COORDINATE_DESCENT_NAIVE",
"COORDINATE_DESCENT"][random.randint(0,1)]
if random.randint(0,1): kwargs['standardize'] = [True, False][random.randint(0,1)]
if random.randint(0,1):
if family == "gaussian": kwargs['link'] = gaussian_links[random.randint(0,2)]
elif family == "binomial": kwargs['link'] = binomial_links[random.randint(0,0)]
elif family == "poisson" : kwargs['link'] = poisson_links[random.randint(0,1)]
elif family == "gamma" : kwargs['link'] = gamma_links[random.randint(0,2)]
if random.randint(0,1): kwargs['alpha'] = [random.random()]
if family == "binomial":
if random.randint(0,1): kwargs['prior'] = random.random()
if random.randint(0,1): kwargs['lambda_search'] = [True, False][random.randint(0,1)]
if 'lambda_search' in kwargs.keys():
if random.randint(0,1): kwargs['nlambdas'] = random.randint(2,10)
do_validation = [True, False][random.randint(0,1)]
# beta constraints
if random.randint(0,1):
bc = []
for n in x:
if train[n].isnumeric():
name = train.names[n]
lower_bound = random.uniform(-1,1)
upper_bound = lower_bound + random.random()
bc.append([name, lower_bound, upper_bound])
if len(bc) > 0:
beta_constraints = h2o.H2OFrame(bc)
beta_constraints.set_names(['names', 'lower_bounds', 'upper_bounds'])
kwargs['beta_constraints'] = beta_constraints.frame_id
# display the parameters and their corresponding values
print "-----------------------"
print "x: {0}".format(x)
print "y: {0}".format(y)
print "validation: {0}".format(do_validation)
for k, v in zip(kwargs.keys(), kwargs.values()):
if k == 'beta_constraints':
print k + ": "
beta_constraints.show()
else:
print k + ": {0}".format(v)
if do_validation:
m = H2OGeneralizedLinearEstimator(**kwargs)
h2o.glm(x=train[x], y=train[y], validation_x=valid[x], validation_y=valid[y], **kwargs)
else: h2o.glm(x=train[x], y=train[y], **kwargs)
print "-----------------------"
def offset_1897():
print "Checking binomial models for GLM with and without offset"
print "Import prostate dataset into H2O and R..."
prostate_hex = h2o.import_file(tests.locate("smalldata/prostate/prostate.csv"))
print "Checking binomial model without offset..."
prostate_glm_h2o = h2o.glm(
x=prostate_hex[["RACE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON"]],
y=prostate_hex["CAPSULE"],
training_frame=prostate_hex,
family="binomial",
standardize=False,
)
print "h2o residual: {0}".format(prostate_glm_h2o.residual_deviance())
print "r residual: {0}".format(379.053509501537)
assert abs(379.053509501537 - prostate_glm_h2o.residual_deviance()) < 0.1
print "Checking binomial model with offset..."
prostate_glm_h2o = h2o.glm(
x=prostate_hex[["RACE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON", "AGE"]],
y=prostate_hex["CAPSULE"],
training_frame=prostate_hex,
family="binomial",
offset_column="AGE",
standardize=False,
)
print "h2o residual: {0}".format(prostate_glm_h2o.residual_deviance())
print "r residual: {0}".format(1515.91815848623)
assert abs(1515.91815848623 - prostate_glm_h2o.residual_deviance()) < 0.1
print "Checking binomial model without offset..."
prostate_glm_h2o = h2o.glm(
x=prostate_hex[["RACE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON"]],
y=prostate_hex["CAPSULE"],
training_frame=prostate_hex,
family="poisson",
standardize=False,
)
print "h2o residual: {0}".format(prostate_glm_h2o.residual_deviance())
print "r residual: {0}".format(216.339989007507)
assert abs(216.339989007507 - prostate_glm_h2o.residual_deviance()) < 0.1
print "Checking binomial model with offset..."
prostate_glm_h2o = h2o.glm(
x=prostate_hex[["RACE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON", "AGE"]],
y=prostate_hex["CAPSULE"],
training_frame=prostate_hex,
family="poisson",
offset_column="AGE",
standardize=False,
)
print "h2o residual: {0}".format(prostate_glm_h2o.residual_deviance())
print "r residual: {0}".format(2761.76218461138)
assert abs(2761.76218461138 - prostate_glm_h2o.residual_deviance()) < 0.1
def attack(family, train, valid, x, y):
kwargs = {}
kwargs['family'] = family
gaussian_links = ["inverse", "log", "identity"]
binomial_links = ["logit"]
poisson_links = ["log", "identity"]
gamma_links = ["inverse", "log", "identity"]
# randomly select parameters and their corresponding values
if random.randint(0,1): kwargs['max_iterations'] = random.randint(1,50)
if random.random() > 0.8: kwargs['beta_epsilon'] = random.random()
if random.randint(0,1): kwargs['solver'] = ["AUTO", "IRLSM", "L_BFGS", "COORDINATE_DESCENT_NAIVE",
"COORDINATE_DESCENT"][random.randint(0,1)]
if random.randint(0,1): kwargs['standardize'] = [True, False][random.randint(0,1)]
if random.randint(0,1):
if family == "gaussian": kwargs['link'] = gaussian_links[random.randint(0,2)]
elif family == "binomial": kwargs['link'] = binomial_links[random.randint(0,0)]
elif family == "poisson" : kwargs['link'] = poisson_links[random.randint(0,1)]
elif family == "gamma" : kwargs['link'] = gamma_links[random.randint(0,2)]
if random.randint(0,1): kwargs['alpha'] = [random.random()]
if family == "binomial":
if random.randint(0,1): kwargs['prior'] = random.random()
if random.randint(0,1): kwargs['lambda_search'] = [True, False][random.randint(0,1)]
if 'lambda_search' in list(kwargs.keys()):
if random.randint(0,1): kwargs['nlambdas'] = random.randint(2,10)
do_validation = [True, False][random.randint(0,1)]
# beta constraints
if random.randint(0,1):
bc = []
for n in x:
if train[n].isnumeric():
name = train.names[n]
lower_bound = random.uniform(-1,1)
upper_bound = lower_bound + random.random()
bc.append([name, lower_bound, upper_bound])
if len(bc) > 0:
beta_constraints = h2o.H2OFrame(bc)
beta_constraints.set_names(['names', 'lower_bounds', 'upper_bounds'])
kwargs['beta_constraints'] = beta_constraints.frame_id
# display the parameters and their corresponding values
print("-----------------------")
print("x: {0}".format(x))
print("y: {0}".format(y))
print("validation: {0}".format(do_validation))
for k, v in zip(list(kwargs.keys()), list(kwargs.values())):
if k == 'beta_constraints':
print(k + ": ")
beta_constraints.show()
else:
print(k + ": {0}".format(v))
if do_validation: h2o.glm(x=train[x], y=train[y], validation_x=valid[x], validation_y=valid[y], **kwargs)
else: h2o.glm(x=train[x], y=train[y], **kwargs)
print("-----------------------")
def check_same(data1, data2):
glm1_regression = h2o.glm(x=data1[2:20], y=data1[1])
glm2_regression = h2o.glm(x=data2[2:21], y=data2[1], weights_column="weights")
glm1_binomial = h2o.glm(x=data1[1:20], y=data1[0], family="binomial")
glm2_binomial = h2o.glm(x=data2[1:21], y=data2[0], weights_column="weights", family="binomial")
assert abs(glm1_regression.mse() - glm2_regression.mse()) < 1e-6, "Expected mse's to be the same, but got {0}, " \
"and {1}".format(glm1_regression.mse(),
glm2_regression.mse())
assert abs(glm1_binomial.auc() - glm2_binomial.auc()) < 1e-6, "Expected auc's to be the same, but got {0}, and " \
"{1}".format(glm1_binomial.auc(), glm2_binomial.auc())
def offset_1897():
print 'Checking binomial models for GLM with and without offset'
print 'Import prostate dataset into H2O and R...'
prostate_hex = h2o.import_file(
pyunit_utils.locate("smalldata/prostate/prostate.csv"))
print "Checking binomial model without offset..."
prostate_glm_h2o = h2o.glm(
x=prostate_hex[["RACE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON"]],
y=prostate_hex["CAPSULE"],
training_frame=prostate_hex,
family="binomial",
standardize=False)
print "h2o residual: {0}".format(prostate_glm_h2o.residual_deviance())
print "r residual: {0}".format(379.053509501537)
assert abs(379.053509501537 - prostate_glm_h2o.residual_deviance()) < 0.1
print "Checking binomial model with offset..."
prostate_glm_h2o = h2o.glm(x=prostate_hex[[
"RACE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON", "AGE"
]],
y=prostate_hex["CAPSULE"],
training_frame=prostate_hex,
family="binomial",
offset_column="AGE",
standardize=False)
print "h2o residual: {0}".format(prostate_glm_h2o.residual_deviance())
print "r residual: {0}".format(1515.91815848623)
assert abs(1515.91815848623 - prostate_glm_h2o.residual_deviance()) < 0.1
print "Checking binomial model without offset..."
prostate_glm_h2o = h2o.glm(
x=prostate_hex[["RACE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON"]],
y=prostate_hex["CAPSULE"],
training_frame=prostate_hex,
family="poisson",
standardize=False)
print "h2o residual: {0}".format(prostate_glm_h2o.residual_deviance())
print "r residual: {0}".format(216.339989007507)
assert abs(216.339989007507 - prostate_glm_h2o.residual_deviance()) < 0.1
print "Checking binomial model with offset..."
prostate_glm_h2o = h2o.glm(x=prostate_hex[[
"RACE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON", "AGE"
]],
y=prostate_hex["CAPSULE"],
training_frame=prostate_hex,
family="poisson",
offset_column="AGE",
standardize=False)
print "h2o residual: {0}".format(prostate_glm_h2o.residual_deviance())
print "r residual: {0}".format(2761.76218461138)
assert abs(2761.76218461138 - prostate_glm_h2o.residual_deviance()) < 0.1
def link_correct_default():
print("Reading in original prostate data.")
h2o_data = h2o.upload_file(path=tests.locate("smalldata/prostate/prostate.csv.zip"))
print("Compare models with link unspecified and canonical link specified.")
print("GAUSSIAN: ")
h2o_model_unspecified = h2o.glm(x=h2o_data[1:8], y=h2o_data[8], family="gaussian")
h2o_model_specified = h2o.glm(x=h2o_data[1:8], y=h2o_data[8], family="gaussian", link="identity")
assert h2o_model_specified._model_json['output']['coefficients_table'].cell_values == \
h2o_model_unspecified._model_json['output']['coefficients_table'].cell_values, "coefficient should be equal"
print("BINOMIAL: ")
h2o_model_unspecified = h2o.glm(x=h2o_data[2:9], y=h2o_data[1], family="binomial")
h2o_model_specified = h2o.glm(x=h2o_data[2:9], y=h2o_data[1], family="binomial", link="logit")
assert h2o_model_specified._model_json['output']['coefficients_table'].cell_values == \
h2o_model_unspecified._model_json['output']['coefficients_table'].cell_values, "coefficient should be equal"
print("POISSON: ")
h2o_model_unspecified = h2o.glm(x=h2o_data[2:9], y=h2o_data[1], family="poisson")
h2o_model_specified = h2o.glm(x=h2o_data[2:9], y=h2o_data[1], family="poisson", link="log")
assert h2o_model_specified._model_json['output']['coefficients_table'].cell_values == \
h2o_model_unspecified._model_json['output']['coefficients_table'].cell_values, "coefficient should be equal"
print("GAMMA: ")
h2o_model_unspecified = h2o.glm(x=h2o_data[3:9], y=h2o_data[2], family="gamma")
h2o_model_specified = h2o.glm(x=h2o_data[3:9], y=h2o_data[2], family="gamma", link="inverse")
assert h2o_model_specified._model_json['output']['coefficients_table'].cell_values == \
h2o_model_unspecified._model_json['output']['coefficients_table'].cell_values, "coefficient should be equal"
def check_same(data1, data2):
glm1_regression = h2o.glm(x=data1[2:20], y=data1[1])
glm2_regression = h2o.glm(x=data2[2:21], y=data2[1], weights_column="weights", training_frame=data2)
glm1_binomial = h2o.glm(x=data1[1:20], y=data1[0], family="binomial")
glm2_binomial = h2o.glm(x=data2[1:21], y=data2[0], weights_column="weights", family="binomial",training_frame=data2)
assert abs(glm1_regression.mse() - glm2_regression.mse()) < 1e-6, "Expected mse's to be the same, but got {0}, " \
"and {1}".format(glm1_regression.mse(),
glm2_regression.mse())
assert abs(glm1_binomial.null_deviance() - glm2_binomial.null_deviance()) < 1e-6, \
"Expected null deviances to be the same, but got {0}, and {1}".format(glm1_binomial.null_deviance(),
glm2_binomial.null_deviance())
assert abs(glm1_binomial.residual_deviance() - glm2_binomial.residual_deviance()) < 1e-6, \
"Expected residual deviances to be the same, but got {0}, and {1}".format(glm1_binomial.residual_deviance(),
glm2_binomial.residual_deviance())
def check_same(data1, data2):
glm1_regression = h2o.glm(x=data1[2:20], y=data1[1])
glm2_regression = h2o.glm(x=data2[2:21], y=data2[1], weights_column="weights", training_frame=data2)
glm1_binomial = h2o.glm(x=data1[1:20], y=data1[0], family="binomial")
glm2_binomial = h2o.glm(x=data2[1:21], y=data2[0], weights_column="weights", family="binomial",training_frame=data2)
assert abs(glm1_regression.mse() - glm2_regression.mse()) < 1e-6, "Expected mse's to be the same, but got {0}, " \
"and {1}".format(glm1_regression.mse(),
glm2_regression.mse())
assert abs(glm1_binomial.null_deviance() - glm2_binomial.null_deviance()) < 1e-6, \
"Expected null deviances to be the same, but got {0}, and {1}".format(glm1_binomial.null_deviance(),
glm2_binomial.null_deviance())
assert abs(glm1_binomial.residual_deviance() - glm2_binomial.residual_deviance()) < 1e-6, \
"Expected residual deviances to be the same, but got {0}, and {1}".format(glm1_binomial.residual_deviance(),
glm2_binomial.residual_deviance())
def pyunit_make_glm_model():
# TODO: PUBDEV-1717
pros = h2o.import_file(tests.locate("smalldata/prostate/prostate.csv"))
model = h2o.glm(x=pros[["AGE", "DPROS", "DCAPS", "PSA", "VOL", "GLEASON"]],
y=pros["CAPSULE"],
family="gaussian",
alpha=[0])
new_betas = {
"AGE": 0.5,
"DPROS": 0.5,
"DCAPS": 0.5,
"PSA": 0.5,
"VOL": 0.5,
"GLEASON": 0.5
}
names = '['
for n in new_betas.keys():
names += "\"" + n + "\","
names = names[0:len(names) - 1] + "]"
betas = '['
for b in new_betas.values():
betas += str(b) + ","
betas = betas[0:len(betas) - 1] + "]"
res = h2o.H2OConnection.post_json("MakeGLMModel",
model=model._id,
names=names,
beta=betas)
def link_functions_tweedie_basic(ip, port):
print "Read in prostate data."
hdf = h2o.upload_file(
h2o.locate("smalldata/prostate/prostate_complete.csv.zip"))
print "Testing for family: TWEEDIE"
print "Set variables for h2o."
y = "CAPSULE"
x = ["AGE", "RACE", "DCAPS", "PSA", "VOL", "DPROS", "GLEASON"]
print "Create models with canonical link: TWEEDIE"
model_h2o_tweedie = h2o.glm(x=hdf[x],
y=hdf[y],
family="tweedie",
link="tweedie",
alpha=[0.5],
Lambda=[0])
print "Compare model deviances for link function tweedie (using precomputed values from R)"
deviance_h2o_tweedie = model_h2o_tweedie.residual_deviance(
) / model_h2o_tweedie.null_deviance()
assert 0.721452 - deviance_h2o_tweedie <= 0.01, "h2o's residual/null deviance is more than 0.01 lower than R's. h2o: " \
"{0}, r: {1}".format(deviance_h2o_tweedie, 0.721452)
def perfectSeparation_balanced(ip, port):
# Connect to h2o
h2o.init(ip, port)
print("Read in synthetic balanced dataset")
data = h2o.import_frame(
path=h2o.locate("smalldata/synthetic_perfect_separation/balanced.csv"))
print("Fit model on dataset")
model = h2o.glm(x=data[["x1", "x2"]],
y=data["y"],
family="binomial",
lambda_search=True,
use_all_factor_levels=True,
alpha=[0.5],
Lambda=[0])
print(
"Extract models' coefficients and assert reasonable values (ie. no greater than 50)"
)
print("Balanced dataset")
coef = [
c[1]
for c in model._model_json['output']['coefficients_table'].cell_values
if c[0] != "Intercept"
]
for c in coef:
assert c < 50, "coefficient is too large"
def link_functions_tweedie_vpow(ip,port):
# Connect to h2o
h2o.init(ip,port)
# Load example data from HDtweedie, y = aggregate claim loss
hdf = h2o.upload_file(h2o.locate("smalldata/glm_test/auto.csv"))
y = "y"
x = list(set(hdf.names()) - set(["y"]))
print "Testing for family: TWEEDIE"
print "Create models with canonical link: TWEEDIE"
# Iterate over different variance powers for tweedie
vpower = [0, 1, 1.5]
r_dev = [0.7516627, 0.6708826, 0.7733762]
r_null = [221051.88369951, 32296.29783702, 20229.47425307]
for ridx, vpow in enumerate(vpower):
print "Fit h2o.glm:"
h2ofit = h2o.glm(x=hdf[x], y=hdf[y], family="tweedie", link="tweedie", tweedie_variance_power=vpow, tweedie_link_power=1-vpow,
alpha=[0.5], Lambda=[0])
print "Testing Tweedie variance power: {0}".format(vpow)
print "Compare model deviances for link function tweedie"
deviance_h2o_tweedie = h2ofit.residual_deviance() / h2ofit.null_deviance()
assert r_dev[ridx] - deviance_h2o_tweedie <= 0.01, "h2o's residual/null deviance is more than 0.01 lower than " \
"R's. h2o: {0}, r: {1}".format(deviance_h2o_tweedie, r_dev[ridx])
print "compare null and residual deviance between R glm and h2o.glm for tweedie"
assert abs(r_null[ridx] - h2ofit.null_deviance()) < 1e-6, "h2o's null deviance is not equal to R's. h2o: {0}, r: " \
"{1}".format(h2ofit.null_deviance(), r_null[ridx])
def pubdev_1953():
# small_test = [h2o.locate("bigdata/laptop/citibike-nyc/2013-10.csv")]
# data = h2o.import_file(path=small_test)
# startime = data["starttime"]
# secsPerDay=1000*60*60*24
# data["Days"] = (startime/secsPerDay).floor()
# grouped = data.group_by(["Days","start station name"])
# bpd = grouped.count(name="bikes").get_frame()
# secs = bpd["Days"]*secsPerDay
# bpd["Month"] = secs.month().asfactor()
# bpd["DayOfWeek"] = secs.dayOfWeek()
# wthr1 = h2o.import_file(path=[h2o.locate("bigdata/laptop/citibike-nyc/31081_New_York_City__Hourly_2013.csv"), h2o.locate("bigdata/laptop/citibike-nyc/31081_New_York_City__Hourly_2014.csv")])
# wthr2 = wthr1[["Year Local","Month Local","Day Local","Hour Local","Dew Point (C)","Humidity Fraction","Precipitation One Hour (mm)","Temperature (C)","Weather Code 1/ Description"]]
# wthr2.set_name(wthr2.index("Precipitation One Hour (mm)"), "Rain (mm)")
# wthr2.set_name(wthr2.index("Weather Code 1/ Description"), "WC1")
# wthr3 = wthr2[ wthr2["Hour Local"]==12 ]
# wthr3["msec"] = h2o.H2OFrame.mktime(year=wthr3["Year Local"], month=wthr3["Month Local"]-1, day=wthr3["Day Local"]-1, hour=wthr3["Hour Local"])
# secsPerDay=1000*60*60*24
# wthr3["Days"] = (wthr3["msec"]/secsPerDay).floor()
# wthr4 = wthr3.drop("Year Local").drop("Month Local").drop("Day Local").drop("Hour Local").drop("msec")
# rain = wthr4["Rain (mm)"]
# rain[ rain.isna() ] = 0
# bpd_with_weather = bpd.merge(wthr4,allLeft=True,allRite=False)
# r = bpd_with_weather['Days'].runif(seed=356964763)
# train = bpd_with_weather[ r < 0.6]
# test = bpd_with_weather[(0.6 <= r) & (r < 0.9)]
predictors = ['DayOfWeek', 'WC1', 'start station name', 'Temperature (C)', 'Days', 'Month', 'Humidity Fraction', 'Rain (mm)', 'Dew Point (C)']
train = h2o.import_file(h2o.locate("smalldata/glm_test/citibike_small_train.csv"))
test = h2o.import_file(h2o.locate("smalldata/glm_test/citibike_small_test.csv"))
glm0 = h2o.glm(x=train[predictors], y=train["bikes"], validation_x=test[predictors], validation_y=test["bikes"], family="poisson")
def wide_dataset_large():
print("Reading in Arcene training data for binomial modeling.")
trainDataResponse = np.genfromtxt(tests.locate("smalldata/arcene/arcene_train_labels.labels"), delimiter=' ')
trainDataResponse = np.where(trainDataResponse == -1, 0, 1)
trainDataFeatures = np.genfromtxt(tests.locate("smalldata/arcene/arcene_train.data"), delimiter=' ')
trainData = h2o.H2OFrame(np.column_stack((trainDataResponse, trainDataFeatures)).tolist())
print("Run model on 3250 columns of Arcene with strong rules off.")
model = h2o.glm(x=trainData[1:3250], y=trainData[0].asfactor(), family="binomial", lambda_search=False, alpha=[1])
print("Test model on validation set.")
validDataResponse = np.genfromtxt(tests.locate("smalldata/arcene/arcene_valid_labels.labels"), delimiter=' ')
validDataResponse = np.where(validDataResponse == -1, 0, 1)
validDataFeatures = np.genfromtxt(tests.locate("smalldata/arcene/arcene_valid.data"), delimiter=' ')
validData = h2o.H2OFrame(np.column_stack((validDataResponse, validDataFeatures)).tolist())
prediction = model.predict(validData)
print("Check performance of predictions.")
performance = model.model_performance(validData)
print("Check that prediction AUC better than guessing (0.5).")
assert performance.auc() > 0.5, "predictions should be better then pure chance"
def prostate(ip, port):
# Connect to h2o
h2o.init(ip, port)
# Log.info("Importing prostate.csv data...\n")
h2o_data = h2o.upload_file(
path=h2o.locate("smalldata/logreg/prostate.csv"))
#prostate.summary()
sm_data = pd.read_csv(
h2o.locate("smalldata/logreg/prostate.csv")).as_matrix()
sm_data_response = sm_data[:, 1]
sm_data_features = sm_data[:, 2:]
#Log.info(cat("B)H2O GLM (binomial) with parameters:\nX:", myX, "\nY:", myY, "\n"))
h2o_glm = h2o.glm(y=h2o_data[1],
x=h2o_data[2:],
family="binomial",
n_folds=10,
alpha=[0.5])
h2o_glm.show()
sm_glm = sm.GLM(endog=sm_data_response,
exog=sm_data_features,
family=sm.families.Binomial()).fit()
assert abs(sm_glm.null_deviance -
h2o_glm._model_json['output']['null_deviance']
) < 1e-5, "Expected null deviances to be the same"
def pubdev_1953():
# small_test = [tests.locate("bigdata/laptop/citibike-nyc/2013-10.csv")]
# data = h2o.import_file(path=small_test)
# startime = data["starttime"]
# secsPerDay=1000*60*60*24
# data["Days"] = (startime/secsPerDay).floor()
# grouped = data.group_by(["Days","start station name"])
# bpd = grouped.count(name="bikes").get_frame()
# secs = bpd["Days"]*secsPerDay
# bpd["Month"] = secs.month().asfactor()
# bpd["DayOfWeek"] = secs.dayOfWeek()
# wthr1 = h2o.import_file(path=[tests.locate("bigdata/laptop/citibike-nyc/31081_New_York_City__Hourly_2013.csv"), tests.locate("bigdata/laptop/citibike-nyc/31081_New_York_City__Hourly_2014.csv")])
# wthr2 = wthr1[["Year Local","Month Local","Day Local","Hour Local","Dew Point (C)","Humidity Fraction","Precipitation One Hour (mm)","Temperature (C)","Weather Code 1/ Description"]]
# wthr2.set_name(wthr2.index("Precipitation One Hour (mm)"), "Rain (mm)")
# wthr2.set_name(wthr2.index("Weather Code 1/ Description"), "WC1")
# wthr3 = wthr2[ wthr2["Hour Local"]==12 ]
# wthr3["msec"] = h2o.H2OFrame.mktime(year=wthr3["Year Local"], month=wthr3["Month Local"]-1, day=wthr3["Day Local"]-1, hour=wthr3["Hour Local"])
# secsPerDay=1000*60*60*24
# wthr3["Days"] = (wthr3["msec"]/secsPerDay).floor()
# wthr4 = wthr3.drop("Year Local").drop("Month Local").drop("Day Local").drop("Hour Local").drop("msec")
# rain = wthr4["Rain (mm)"]
# rain[ rain.isna() ] = 0
# bpd_with_weather = bpd.merge(wthr4,allLeft=True,allRite=False)
# r = bpd_with_weather['Days'].runif(seed=356964763)
# train = bpd_with_weather[ r < 0.6]
# test = bpd_with_weather[(0.6 <= r) & (r < 0.9)]
predictors = ['DayOfWeek', 'WC1', 'start station name', 'Temperature (C)', 'Days', 'Month', 'Humidity Fraction', 'Rain (mm)', 'Dew Point (C)']
train = h2o.import_file(tests.locate("smalldata/glm_test/citibike_small_train.csv"))
test = h2o.import_file(tests.locate("smalldata/glm_test/citibike_small_test.csv"))
glm0 = h2o.glm(x=train[predictors], y=train["bikes"], validation_x=test[predictors], validation_y=test["bikes"], family="poisson")
def perfectSeparation_unbalanced():
print("Read in synthetic unbalanced dataset")
data = h2o.import_file(
tests.locate("smalldata/synthetic_perfect_separation/unbalanced.csv"))
print("Fit model on dataset.")
model = h2o.glm(x=data[["x1", "x2"]],
y=data["y"],
family="binomial",
lambda_search=True,
alpha=[0.5],
Lambda=[0])
print(
"Extract models' coefficients and assert reasonable values (ie. no greater than 50)"
)
print("Unbalanced dataset")
coef = [
c[1]
for c in model._model_json['output']['coefficients_table'].cell_values
if c[0] != "Intercept"
]
for c in coef:
assert c < 50, "coefficient is too large"
def save_load_model():
prostate = h2o.import_file(tests.locate("smalldata/prostate/prostate.csv"))
prostate["CAPSULE"] = prostate["CAPSULE"].asfactor()
prostate_glm = h2o.glm(y=prostate["CAPSULE"],
x=prostate[["AGE", "RACE", "PSA", "DCAPS"]],
family="binomial",
alpha=[0.5])
path = os.path.normpath(
os.path.join(os.path.dirname(os.path.realpath(__file__)), "..",
"results"))
assert os.path.isdir(
path), "Expected save directory {0} to exist, but it does not.".format(
path)
model_path = h2o.save_model(prostate_glm, path=path, force=True)
assert os.path.isdir(
model_path
), "Expected load directory {0} to exist, but it does not.".format(
model_path)
the_model = h2o.load_model(model_path)
assert isinstance(
the_model,
H2OBinomialModel), "Expected and H2OBinomialModel, but got {0}".format(
the_model)
def link_functions_binomial():
print("Read in prostate data.")
h2o_data = h2o.import_file(path=tests.locate("smalldata/prostate/prostate_complete.csv.zip"))
h2o_data.head()
sm_data = pd.read_csv(zipfile.ZipFile(tests.locate("smalldata/prostate/prostate_complete.csv.zip")).open("prostate_complete.csv")).as_matrix()
sm_data_response = sm_data[:,2]
sm_data_features = sm_data[:,[1,3,4,5,6,7,8,9]]
print("Testing for family: BINOMIAL")
print("Set variables for h2o.")
myY = "CAPSULE"
myX = ["ID","AGE","RACE","GLEASON","DCAPS","PSA","VOL","DPROS"]
print("Create models with canonical link: LOGIT")
h2o_model = h2o.glm(x=h2o_data[myX], y=h2o_data[myY].asfactor(), family="binomial", link="logit",alpha=[0.5], Lambda=[0])
sm_model = sm.GLM(endog=sm_data_response, exog=sm_data_features, family=sm.families.Binomial(sm.families.links.logit)).fit()
print("Compare model deviances for link function logit")
h2o_deviance = h2o_model.residual_deviance() / h2o_model.null_deviance()
sm_deviance = sm_model.deviance / sm_model.null_deviance
assert h2o_deviance - sm_deviance < 0.01, "expected h2o to have an equivalent or better deviance measures"
def benign(ip, port):
# Connect to h2o
h2o.init(ip, port)
training_data = h2o.import_frame(h2o.locate("smalldata/logreg/benign.csv"))
Y = 3
X = range(3) + range(4, 11)
#Log.info("Build the model")
model = h2o.glm(y=training_data[Y].asfactor(),
x=training_data[X],
family="binomial",
alpha=[0],
Lambda=[1e-5])
#Log.info("Check that the columns used in the model are the ones we passed in.")
#Log.info("===================Columns passed in: ================")
in_names = [training_data.names()[i] for i in X]
#Log.info("===================Columns passed out: ================")
out_names = [
model._model_json['output']['coefficients_table'].cell_values[c][0]
for c in range(len(X) + 1)
]
assert in_names == out_names[1:]
def link_functions_gaussian(ip,port):
print("Read in prostate data.")
h2o_data = h2o.import_file(path=h2o.locate("smalldata/prostate/prostate_complete.csv.zip"))
h2o_data.head()
sm_data = pd.read_csv(zipfile.ZipFile(h2o.locate("smalldata/prostate/prostate_complete.csv.zip")).
open("prostate_complete.csv")).as_matrix()
sm_data_response = sm_data[:,9]
sm_data_features = sm_data[:,1:9]
print("Testing for family: GAUSSIAN")
print("Set variables for h2o.")
myY = "GLEASON"
myX = ["ID","AGE","RACE","CAPSULE","DCAPS","PSA","VOL","DPROS"]
print("Create models with canonical link: IDENTITY")
h2o_model = h2o.glm(x=h2o_data[myX], y=h2o_data[myY], family="gaussian", link="identity",alpha=[0.5], Lambda=[0])
sm_model = sm.GLM(endog=sm_data_response, exog=sm_data_features,
family=sm.families.Gaussian(sm.families.links.identity)).fit()
print("Compare model deviances for link function identity")
h2o_deviance = h2o_model.residual_deviance() / h2o_model.null_deviance()
sm_deviance = sm_model.deviance / sm_model.null_deviance
assert h2o_deviance - sm_deviance < 0.01, "expected h2o to have an equivalent or better deviance measures"
def link_functions_gaussian():
print("Read in prostate data.")
h2o_data = h2o.import_file(
path=tests.locate("smalldata/prostate/prostate_complete.csv.zip"))
h2o_data.head()
sm_data = pd.read_csv(
zipfile.ZipFile(
tests.locate("smalldata/prostate/prostate_complete.csv.zip")).open(
"prostate_complete.csv")).as_matrix()
sm_data_response = sm_data[:, 9]
sm_data_features = sm_data[:, 1:9]
print("Testing for family: GAUSSIAN")
print("Set variables for h2o.")
myY = "GLEASON"
myX = ["ID", "AGE", "RACE", "CAPSULE", "DCAPS", "PSA", "VOL", "DPROS"]
print("Create models with canonical link: IDENTITY")
h2o_model = h2o.glm(x=h2o_data[myX],
y=h2o_data[myY],
family="gaussian",
link="identity",
alpha=[0.5],
Lambda=[0])
sm_model = sm.GLM(endog=sm_data_response,
exog=sm_data_features,
family=sm.families.Gaussian(
sm.families.links.identity)).fit()
print("Compare model deviances for link function identity")
h2o_deviance = h2o_model.residual_deviance() / h2o_model.null_deviance()
sm_deviance = sm_model.deviance / sm_model.null_deviance
assert h2o_deviance - sm_deviance < 0.01, "expected h2o to have an equivalent or better deviance measures"
def glm_solvers():
training_data = h2o.import_file(
pyunit_utils.locate("smalldata/junit/cars_20mpg.csv"))
predictors = ["displacement", "power", "weight", "acceleration", "year"]
for solver in [
"AUTO", "IRLSM", "L_BFGS", "COORDINATE_DESCENT_NAIVE",
"COORDINATE_DESCENT"
]:
print "Solver = {0}".format(solver)
for family in ["binomial", "gaussian", "poisson", "tweedie", "gamma"]:
if family == 'binomial': response_col = "economy_20mpg"
elif family == 'gaussian': response_col = "economy"
else: response_col = "cylinders"
print "Family = {0}".format(family)
if family == 'binomial':
training_data[response_col] = training_data[
response_col].asfactor()
else:
training_data[response_col] = training_data[
response_col].asnumeric()
model = h2o.glm(x=training_data[predictors],
y=training_data[response_col],
family=family,
alpha=[0],
Lambda=[1e-5],
solver=solver)
def wide_dataset_large(ip,port):
# Connect to h2o
h2o.init(ip,port)
print("Reading in Arcene training data for binomial modeling.")
trainDataResponse = np.genfromtxt(h2o.locate("smalldata/arcene/arcene_train_labels.labels"), delimiter=' ')
trainDataResponse = np.where(trainDataResponse == -1, 0, 1)
trainDataFeatures = np.genfromtxt(h2o.locate("smalldata/arcene/arcene_train.data"), delimiter=' ')
trainData = h2o.H2OFrame(np.column_stack((trainDataResponse, trainDataFeatures)).tolist())
print("Run model on 3250 columns of Arcene with strong rules off.")
model = h2o.glm(x=trainData[1:3250], y=trainData[0].asfactor(), family="binomial", lambda_search=False, alpha=[1], use_all_factor_levels=True)
print("Test model on validation set.")
validDataResponse = np.genfromtxt(h2o.locate("smalldata/arcene/arcene_valid_labels.labels"), delimiter=' ')
validDataResponse = np.where(validDataResponse == -1, 0, 1)
validDataFeatures = np.genfromtxt(h2o.locate("smalldata/arcene/arcene_valid.data"), delimiter=' ')
validData = h2o.H2OFrame(np.column_stack((validDataResponse, validDataFeatures)).tolist())
prediction = model.predict(validData)
print("Check performance of predictions.")
performance = model.model_performance(validData)
print("Check that prediction AUC better than guessing (0.5).")
assert performance.auc() > 0.5, "predictions should be better then pure chance"
def link_functions_binomial(ip,port):
# Connect to h2o
h2o.init(ip,port)
print("Read in prostate data.")
h2o_data = h2o.import_frame(path=h2o.locate("smalldata/prostate/prostate_complete.csv.zip"))
h2o_data.head()
sm_data = pd.read_csv(zipfile.ZipFile(h2o.locate("smalldata/prostate/prostate_complete.csv.zip")).open("prostate_complete.csv")).as_matrix()
sm_data_response = sm_data[:,2]
sm_data_features = sm_data[:,[1,3,4,5,6,7,8,9]]
print("Testing for family: BINOMIAL")
print("Set variables for h2o.")
myY = "CAPSULE"
myX = ["ID","AGE","RACE","GLEASON","DCAPS","PSA","VOL","DPROS"]
print("Create models with canonical link: LOGIT")
h2o_model = h2o.glm(x=h2o_data[myX], y=h2o_data[myY].asfactor(), family="binomial", link="logit",alpha=[0.5], Lambda=[0])
sm_model = sm.GLM(endog=sm_data_response, exog=sm_data_features, family=sm.families.Binomial(sm.families.links.logit)).fit()
print("Compare model deviances for link function logit")
h2o_deviance = h2o_model._model_json['output']['residual_deviance'] / h2o_model._model_json['output']['null_deviance']
sm_deviance = sm_model.deviance / sm_model.null_deviance
assert h2o_deviance - sm_deviance < 0.01, "expected h2o to have an equivalent or better deviance measures"
def link_functions_tweedie_vpow():
# Load example data from HDtweedie, y = aggregate claim loss
hdf = h2o.upload_file(pyunit_utils.locate("smalldata/glm_test/auto.csv"))
y = "y"
x = list(set(hdf.names) - set(["y"]))
print "Testing for family: TWEEDIE"
print "Create models with canonical link: TWEEDIE"
# Iterate over different variance powers for tweedie
vpower = [0, 1, 1.5]
r_dev = [0.7516627, 0.6708826, 0.7733762]
r_null = [221051.88369951, 32296.29783702, 20229.47425307]
for ridx, vpow in enumerate(vpower):
print "Fit h2o.glm:"
h2ofit = h2o.glm(x=hdf[x], y=hdf[y], family="tweedie", link="tweedie", tweedie_variance_power=vpow, tweedie_link_power=1-vpow,
alpha=[0.5], Lambda=[0])
print "Testing Tweedie variance power: {0}".format(vpow)
print "Compare model deviances for link function tweedie"
deviance_h2o_tweedie = h2ofit.residual_deviance() / h2ofit.null_deviance()
assert r_dev[ridx] - deviance_h2o_tweedie <= 0.01, "h2o's residual/null deviance is more than 0.01 lower than " \
"R's. h2o: {0}, r: {1}".format(deviance_h2o_tweedie, r_dev[ridx])
print "compare null and residual deviance between R glm and h2o.glm for tweedie"
assert abs(r_null[ridx] - h2ofit.null_deviance()) < 1e-6, "h2o's null deviance is not equal to R's. h2o: {0}, r: " \
"{1}".format(h2ofit.null_deviance(), r_null[ridx])
def grid_lambda_search():
# Log.info("Importing prostate.csv data...\n")
prostate = h2o.import_file(path=pyunit_utils.locate("smalldata/logreg/prostate.csv"))
#prostate.summary()
# Log.info("H2O GLM (binomial) with parameters: alpha = c(0.25, 0.5), nlambda = 20, lambda_search = TRUE, nfolds: 2\n")
model = h2o.glm(x=prostate[2:9], y=prostate[1], family="binomial", nlambdas=5, lambda_search=True, n_folds=2)
if random.random() < 0.5:
model_idx = 0
else:
model_idx = 1
model_bestlambda = model.models(model_idx)
params_bestlambda = model.params()
# Log.info(cat("All lambda values returned:\n", params_bestlambda.lambdas()))
assert len(params_bestlambda.lambdas()) <= 5, "expected 5 or less lambdas"
random_lambda = random.choice(params_bestlambda.lambdas())
print("RANDOM LAMBDA")
print(random_lambda)
# Log.info(cat("Retrieving model corresponding to alpha =", params_bestlambda.alpha(), "and randomly chosen lambda", random_lambda, "\n"))
random_model = model.getGLMLambdaModel(model_bestlambda, random_lambda)
# Log.info("EXPECTING THESE TO BE EQUAL")
print(random_model.Lambda())
print(random_lambda)
assert random_model.Lambda() == random_lambda, "expected lambdas to be equal"
# Log.info(cat("Retrieving model corresponding to alpha =", params_bestlambda.alpha(), "and best lambda", params_bestlambda.lambdaBest(), "\n"))
best_model = h2o.getGLMLambdaModel(model_bestlambda, params_bestlambda.lambda_best())
assert best_model.model() == model_bestlambda.model(), "expected models to be equal"
# Log.info("H2O GLM (binomial) with parameters: alpha = [0.25, 0.5], nlambda = 20, lambda_search = TRUE, nfolds: 2\n")
prostate_search = h2o.glm(x=prostate[2:9], y=prostate[1], family="binomial", alpha=[0.25, 0.5], nlambdas=5, lambda_search=True, n_folds=2)
model_search = prostate_search.models(model_idx)
models_best = model_search.models(model_search.best_model())
params_best = models_best.params()
assert params_bestlambda.lambda_best() == params_best.lambda_best(), "expected lambdas to be equal"
assert len(params_best.lambda_all()) <= 20, "expected 20 or fewer lambdas"
def link_functions_poisson():
print("Read in prostate data.")
h2o_data = h2o.import_file(path=pyunit_utils.locate("smalldata/prostate/prostate_complete.csv.zip"))
sm_data = pd.read_csv(
zipfile.ZipFile(pyunit_utils.locate("smalldata/prostate/prostate_complete.csv.zip")).open(
"prostate_complete.csv"
)
).as_matrix()
sm_data_response = sm_data[:, 9]
sm_data_features = sm_data[:, 1:9]
print("Testing for family: POISSON")
print("Set variables for h2o.")
myY = "GLEASON"
myX = ["ID", "AGE", "RACE", "CAPSULE", "DCAPS", "PSA", "VOL", "DPROS"]
print("Create h2o model with canonical link: LOG")
h2o_model_log = h2o.glm(x=h2o_data[myX], y=h2o_data[myY], family="poisson", link="log", alpha=[0.5], Lambda=[0])
print("Create statsmodel model with canonical link: LOG")
sm_model_log = sm.GLM(
endog=sm_data_response, exog=sm_data_features, family=sm.families.Poisson(sm.families.links.log)
).fit()
print("Compare model deviances for link function log")
h2o_deviance_log = old_div(h2o_model_log.residual_deviance(), h2o_model_log.null_deviance())
sm_deviance_log = old_div(sm_model_log.deviance, sm_model_log.null_deviance)
assert h2o_deviance_log - sm_deviance_log < 0.01, "expected h2o to have an equivalent or better deviance measures"
print("Create h2o models with link: IDENTITY")
h2o_model_id = h2o.glm(x=h2o_data[myX], y=h2o_data[myY], family="poisson", link="identity", alpha=[0.5], Lambda=[0])
print("Create statsmodel models with link: IDENTITY")
sm_model_id = sm.GLM(
endog=sm_data_response, exog=sm_data_features, family=sm.families.Poisson(sm.families.links.identity)
).fit()
print("Compare model deviances for link function identity")
h2o_deviance_id = old_div(h2o_model_id.residual_deviance(), h2o_model_id.null_deviance())
sm_deviance_id = old_div(sm_model_id.deviance, sm_model_id.null_deviance)
assert h2o_deviance_id - sm_deviance_id < 0.01, "expected h2o to have an equivalent or better deviance measures"
def pubdev_1839():
train = h2o.import_file(pyunit_utils.locate("smalldata/jira/pubdev_1839_repro_train.csv"))
test = h2o.import_file(pyunit_utils.locate("smalldata/jira/pubdev_1839_repro_test.csv"))
glm0 = h2o.glm(x =train.drop("bikes"),
y =train ["bikes"],
validation_x=test .drop("bikes"),
validation_y=test ["bikes"],
family="poisson")
def pubdev_1839(ip, port):
train = h2o.import_file(h2o.locate("smalldata/jira/pubdev_1839_repro_train.csv"))
test = h2o.import_file(h2o.locate("smalldata/jira/pubdev_1839_repro_test.csv"))
glm0 = h2o.glm(x =train.drop("bikes"),
y =train ["bikes"],
validation_x=test .drop("bikes"),
validation_y=test ["bikes"],
Lambda=[1e-5],
family="poisson")
def pubdev_1839(ip, port):
train = h2o.import_file(h2o.locate("smalldata/jira/pubdev_1839_repro_train.csv"))
test = h2o.import_file(h2o.locate("smalldata/jira/pubdev_1839_repro_test.csv"))
glm0 = h2o.glm(x =train.drop("bikes"),
y =train ["bikes"],
validation_x=test .drop("bikes"),
validation_y=test ["bikes"],
Lambda=[1e-5],
family="poisson")
def pubdev_1839():
train = h2o.import_file(
pyunit_utils.locate("smalldata/jira/pubdev_1839_repro_train.csv"))
test = h2o.import_file(
pyunit_utils.locate("smalldata/jira/pubdev_1839_repro_test.csv"))
glm0 = h2o.glm(x=train.drop("bikes"),
y=train["bikes"],
validation_x=test.drop("bikes"),
validation_y=test["bikes"],
family="poisson")
def save_load_model(ip,port):
# Connect to h2o
h2o.init(ip,port)
prostate = h2o.import_frame(h2o.locate("smalldata/prostate/prostate.csv"))
prostate["CAPSULE"] = prostate["CAPSULE"].asfactor()
prostate_glm = h2o.glm(y=prostate["CAPSULE"], x=prostate[["AGE","RACE","PSA","DCAPS"]], family = "binomial",
alpha = [0.5])
model_path = h2o.save_model(prostate_glm, name="delete_model", force=True)
the_model = h2o.load_model(model_path)
assert isinstance(the_model, H2OBinomialModel), "Expected and H2OBinomialModel, but got {0}".format(the_model)
def save_load_model():
prostate = h2o.import_file(h2o.locate("smalldata/prostate/prostate.csv"))
prostate["CAPSULE"] = prostate["CAPSULE"].asfactor()
prostate_glm = h2o.glm(y=prostate["CAPSULE"], x=prostate[["AGE","RACE","PSA","DCAPS"]], family = "binomial",
alpha = [0.5])
model_path = h2o.save_model(prostate_glm,force=True)
the_model = h2o.load_model(model_path)
shutil.rmtree(model_path)
assert isinstance(the_model, H2OBinomialModel), "Expected and H2OBinomialModel, but got {0}".format(the_model)
def save_load_model(ip,port):
prostate = h2o.import_file(h2o.locate("smalldata/prostate/prostate.csv"))
prostate["CAPSULE"] = prostate["CAPSULE"].asfactor()
prostate_glm = h2o.glm(y=prostate["CAPSULE"], x=prostate[["AGE","RACE","PSA","DCAPS"]], family = "binomial",
alpha = [0.5])
model_path = h2o.save_model(prostate_glm,force=True)
the_model = h2o.load_model(model_path)
shutil.rmtree(model_path)
assert isinstance(the_model, H2OBinomialModel), "Expected and H2OBinomialModel, but got {0}".format(the_model)
def pyunit_make_glm_model():
# TODO: PUBDEV-1717
pros = h2o.import_file(tests.locate("smalldata/prostate/prostate.csv"))
model = h2o.glm(x=pros[["AGE","DPROS","DCAPS","PSA","VOL","GLEASON"]], y=pros["CAPSULE"], family="gaussian", alpha=[0])
new_betas = {"AGE":0.5, "DPROS":0.5, "DCAPS":0.5, "PSA":0.5, "VOL":0.5, "GLEASON":0.5}
names = '['
for n in new_betas.keys(): names += "\""+n+"\","
names = names[0:len(names)-1]+"]"
betas = '['
for b in new_betas.values(): betas += str(b)+","
betas = betas[0:len(betas)-1]+"]"
res = h2o.H2OConnection.post_json("MakeGLMModel",model=model._id,names=names,beta=betas)
def prostate():
h2o_data = h2o.upload_file(path=pyunit_utils.locate("smalldata/logreg/prostate.csv"))
h2o_data.summary()
sm_data = pd.read_csv(pyunit_utils.locate("smalldata/logreg/prostate.csv")).as_matrix()
sm_data_response = sm_data[:, 1]
sm_data_features = sm_data[:, 2:]
h2o_glm = h2o.glm(y=h2o_data[1], x=h2o_data[2:], family="binomial", nfolds=10, alpha=[0.5])
sm_glm = sm.GLM(endog=sm_data_response, exog=sm_data_features, family=sm.families.Binomial()).fit()
print "statsmodels null deviance {0}".format(sm_glm.null_deviance)
print "h2o null deviance {0}".format(h2o_glm.null_deviance())
assert abs(sm_glm.null_deviance - h2o_glm.null_deviance()) < 1e-5, "Expected null deviances to be the same"
def glm_solvers():
predictors = ["displacement","power","weight","acceleration","year"]
for solver in ["AUTO", "IRLSM", "L_BFGS", "COORDINATE_DESCENT_NAIVE", "COORDINATE_DESCENT"]:
print("Solver = {0}".format(solver))
for family in ["binomial", "gaussian", "poisson", "tweedie", "gamma"]:
if family == 'binomial': response_col = "economy_20mpg"
elif family == 'gaussian': response_col = "economy"
else: response_col = "cylinders"
print("Family = {0}".format(family))
training_data = h2o.import_file(pyunit_utils.locate("smalldata/junit/cars_20mpg.csv"))
if family == 'binomial': training_data[response_col] = training_data[response_col].asfactor()
else: training_data[response_col] = training_data[response_col].asnumeric()
model = h2o.glm(x=training_data[predictors], y=training_data[response_col], family=family, alpha=[0], Lambda=[1e-5], solver=solver)
h2o.remove(training_data)
def perfectSeparation_unbalanced():
print("Read in synthetic unbalanced dataset")
data = h2o.import_file(pyunit_utils.locate("smalldata/synthetic_perfect_separation/unbalanced.csv"))
print("Fit model on dataset.")
model = h2o.glm(x=data[["x1", "x2"]], y=data["y"], family="binomial", lambda_search=True, alpha=[0.5], Lambda=[0])
print("Extract models' coefficients and assert reasonable values (ie. no greater than 50)")
print("Unbalanced dataset")
coef = [c[1] for c in model._model_json['output']['coefficients_table'].cell_values if c[0] != "Intercept"]
for c in coef:
assert c < 50, "coefficient is too large"
def prostate():
h2o_data = h2o.upload_file(path=tests.locate("smalldata/logreg/prostate.csv"))
h2o_data.summary()
sm_data = pd.read_csv(tests.locate("smalldata/logreg/prostate.csv")).as_matrix()
sm_data_response = sm_data[:,1]
sm_data_features = sm_data[:,2:]
h2o_glm = h2o.glm(y=h2o_data[1], x=h2o_data[2:], family="binomial", nfolds=10, alpha=[0.5])
sm_glm = sm.GLM(endog=sm_data_response, exog=sm_data_features, family=sm.families.Binomial()).fit()
print "statsmodels null deviance {0}".format(sm_glm.null_deviance)
print "h2o null deviance {0}".format(h2o_glm.null_deviance())
assert abs(sm_glm.null_deviance - h2o_glm.null_deviance()) < 1e-5, "Expected null deviances to be the same"
def save_load_model():
prostate = h2o.import_file(h2o.locate("smalldata/prostate/prostate.csv"))
prostate["CAPSULE"] = prostate["CAPSULE"].asfactor()
prostate_glm = h2o.glm(y=prostate["CAPSULE"], x=prostate[["AGE","RACE","PSA","DCAPS"]], family = "binomial",
alpha = [0.5])
path = os.path.normpath(os.path.join(os.path.dirname(os.path.realpath(__file__)),"..","results"))
assert os.path.isdir(path), "Expected save directory {0} to exist, but it does not.".format(path)
model_path = h2o.save_model(prostate_glm, path=path, force=True)
assert os.path.isdir(model_path), "Expected load directory {0} to exist, but it does not.".format(model_path)
the_model = h2o.load_model(model_path)
assert isinstance(the_model, H2OBinomialModel), "Expected and H2OBinomialModel, but got {0}".format(the_model)
def perfectSeparation_balanced(ip,port):
# Connect to h2o
h2o.init(ip,port)
print("Read in synthetic balanced dataset")
data = h2o.import_frame(path=h2o.locate("smalldata/synthetic_perfect_separation/balanced.csv"))
print("Fit model on dataset")
model = h2o.glm(x=data[["x1", "x2"]], y=data["y"], family="binomial", lambda_search=True, use_all_factor_levels=True, alpha=[0.5], Lambda=[0])
print("Extract models' coefficients and assert reasonable values (ie. no greater than 50)")
print("Balanced dataset")
coef = [c[1] for c in model._model_json['output']['coefficients_table'].cell_values if c[0] != "Intercept"]
for c in coef:
assert c < 50, "coefficient is too large"
def benign():
training_data = h2o.import_file(tests.locate("smalldata/logreg/benign.csv"))
Y = 3
X = range(3) + range(4,11)
#Log.info("Build the model")
model = h2o.glm(y=training_data[Y].asfactor(), x=training_data[X], family="binomial", alpha=[0], Lambda=[1e-5])
#Log.info("Check that the columns used in the model are the ones we passed in.")
#Log.info("===================Columns passed in: ================")
in_names = [training_data.names[i] for i in X]
#Log.info("===================Columns passed out: ================")
out_names = [model._model_json['output']['coefficients_table'].cell_values[c][0] for c in range(len(X)+1)]
assert in_names == out_names[1:]
def glm_mean_residual_deviance(ip,port):
cars = h2o.import_frame(path=h2o.locate("smalldata/junit/cars_20mpg.csv"))
s = cars[0].runif()
train = cars[s > 0.2]
valid = cars[s <= 0.2]
predictors = ["displacement","power","weight","acceleration","year"]
response_col = "economy"
glm = h2o.glm(x=train[predictors],
y=train[response_col],
validation_x=valid[predictors],
validation_y=valid[response_col],
nfolds=3)
glm_mrd = glm.mean_residual_deviance(train=True,valid=True,xval=True)
assert isinstance(glm_mrd['train'],float), "Expected training mean residual deviance to be a float, but got " \
"{0}".format(type(glm_mrd['train']))
assert isinstance(glm_mrd['valid'],float), "Expected validation mean residual deviance to be a float, but got " \
"{0}".format(type(glm_mrd['valid']))
assert isinstance(glm_mrd['xval'],float), "Expected cross-validation mean residual deviance to be a float, but got " \
"{0}".format(type(glm_mrd['xval']))
