def test_GLM2_tweedie(self):
csvFilename = "AutoClaim.csv"
csvPathname = 'standard/' + csvFilename
print "\nStarting", csvPathname
parseResult = h2i.import_parse(bucket='home-0xdiag-datasets', path=csvPathname, schema='put')
# columns start at 0
# regress: glm(CLM_AMT ~ CAR_USE + REVOLKED + GENDER + AREA + MARRIED + CAR_TYPE, data=AutoClaim, family=tweedie(1.34))
coefs = [7, 13, 20, 27, 21, 11]
y = 4
ignored_cols = h2o_cmd.createIgnoredCols(key=parseResult['destination_key'], cols=coefs, response=y)
# sapply(c('CLM_AMT', 'CAR_USE', 'REVOLKED', 'GENDER', 'AREA', 'MARRIED', 'CAR_TYPE'), function(x) which(x==colnames(AutoClaim)) - 1)
kwargs = {
'family': 'tweedie',
'tweedie_variance_power': 1.36,
'response': y,
'ignored_cols' : ignored_cols,
'max_iter': 10,
'lambda': 0,
'alpha': 0,
'n_folds': 0,
'beta_epsilon': 1e-4,
}
glm = h2o_cmd.runGLM(parseResult=parseResult, timeoutSecs=15, **kwargs)
coefficientsExpected = {'Intercept': 0, 'GENDER.M': 0.0014842488782470984, 'CAR_TYPE.Sports Car': 0.07786742314454961, 'MARRIED.Yes': 0.0007748552195851079, 'CAR_TYPE.SUV': 0.07267702940249621, 'CAR_TYPE.Pickup': 0.04952083408742968, 'CAR_TYPE.Van': 0.026422137690691405, 'CAR_TYPE.Sedan': 0.05128350794060489, 'CAR_USE.Private': -0.03050194832853935, 'REVOLKED.Yes': -0.05095942737408699}
deltaExpected = 0.05
(warnings, coefficients, intercept) = h2o_glm.simpleCheckGLM(self, glm, None,
coefficientsExpected=coefficientsExpected, deltaExpected=deltaExpected, **kwargs)
print 'coefficients: %s' % (str(coefficients))
def test_GLM2_umass(self):
h2o.beta_features = True
csvFilenameList = [
('cgd.dat', 'gaussian', 12, 30, None),
('chdage.dat', 'binomial', 2, 30, None),
# leave out ID and birth weight
('clslowbwt.dat', 'binomial', 7, 60, [1, 2, 3, 4, 5]),
('icu.dat', 'binomial', 1, 60, None),
# need to exclude col 0 (ID) and col 10 (bwt)
# but -x doesn't work..so do 2:9...range doesn't work? FIX!
('lowbwt.dat', 'binomial', 1, 60, [2, 3, 4, 5, 6, 7, 8, 9]),
('lowbwtm11.dat', 'binomial', 1, 60, None),
('meexp.dat', 'gaussian', 3, 60, None),
('nhanes3.dat', 'binomial', 15, 60, None),
('pbc.dat', 'gaussian', 1, 60, None),
('pharynx.dat', 'gaussian', 12, 60, None),
('pros.dat', 'binomial', 1, 60, None),
('uis.dat', 'binomial', 8, 60, None),
]
trial = 0
for i in range(3):
for (csvFilename, family, y, timeoutSecs, x) in csvFilenameList:
csvPathname = "logreg/umass_statdata/" + csvFilename
kwargs = {
'n_folds': 3,
'response': y,
'family': family,
'alpha': 1,
'lambda': 1e-4
}
parseResult = h2i.import_parse(bucket='smalldata',
path=csvPathname,
schema='put',
timeoutSecs=timeoutSecs)
if x is not None:
ignored_cols = h2o_cmd.createIgnoredCols(
key=parseResult['destination_key'], cols=x, response=y)
kwargs['ignored_cols'] = ignored_cols
start = time.time()
glm = h2o_cmd.runGLM(parseResult=parseResult,
timeoutSecs=timeoutSecs,
**kwargs)
h2o_glm.simpleCheckGLM(self, glm, None, **kwargs)
print "glm end (w/check) on ", csvPathname, 'took', time.time(
) - start, 'seconds'
trial += 1
print "\nTrial #", trial
def test_GLM2_umass(self):
h2o.beta_features = True
csvFilenameList = [
('cgd.dat', 'gaussian', 12, 5, None),
('chdage.dat', 'binomial', 2, 5, None),
# leave out ID and birth weight
('clslowbwt.dat', 'binomial', 7, 30, [1,2,3,4,5]),
('icu.dat', 'binomial', 1, 30, None),
# need to exclude col 0 (ID) and col 10 (bwt)
# but -x doesn't work..so do 2:9...range doesn't work? FIX!
('lowbwt.dat', 'binomial', 1, 30, [2,3,4,5,6,7,8,9]),
('lowbwtm11.dat', 'binomial', 1, 30, None),
('meexp.dat', 'gaussian', 3, 30, None),
('nhanes3.dat', 'binomial', 15, 30, None),
('pbc.dat', 'gaussian', 1, 30, None),
('pharynx.dat', 'gaussian', 12, 30, None),
('pros.dat', 'binomial', 1, 30, None),
('uis.dat', 'binomial', 8, 30, None),
]
trial = 0
for i in range(3):
for (csvFilename, family, y, timeoutSecs, x) in csvFilenameList:
csvPathname = "logreg/umass_statdata/" + csvFilename
kwargs = {'n_folds': 3, 'response': y, 'family': family, 'alpha': 1, 'lambda': 1e-4}
parseResult = h2i.import_parse(bucket='smalldata', path=csvPathname, schema='put',
timeoutSecs=timeoutSecs)
if x is not None:
ignored_cols = h2o_cmd.createIgnoredCols(key=parseResult['destination_key'],
cols=x, response=y)
kwargs['ignored_cols'] = ignored_cols
start = time.time()
glm = h2o_cmd.runGLM(parseResult=parseResult, timeoutSecs=timeoutSecs, **kwargs)
h2o_glm.simpleCheckGLM(self, glm, None, **kwargs)
print "glm end (w/check) on ", csvPathname, 'took', time.time() - start, 'seconds'
trial += 1
print "\nTrial #", trial
def test_GLM2_umass(self):
h2o.beta_features = True
csvFilenameList = [
("cgd.dat", "gaussian", 12, 5, None),
("chdage.dat", "binomial", 2, 5, None),
# leave out ID and birth weight
("clslowbwt.dat", "binomial", 7, 10, [1, 2, 3, 4, 5]),
("icu.dat", "binomial", 1, 10, None),
# need to exclude col 0 (ID) and col 10 (bwt)
# but -x doesn't work..so do 2:9...range doesn't work? FIX!
("lowbwt.dat", "binomial", 1, 10, [2, 3, 4, 5, 6, 7, 8, 9]),
("lowbwtm11.dat", "binomial", 1, 10, None),
("meexp.dat", "gaussian", 3, 10, None),
("nhanes3.dat", "binomial", 15, 10, None),
("pbc.dat", "gaussian", 1, 10, None),
("pharynx.dat", "gaussian", 12, 10, None),
("pros.dat", "binomial", 1, 10, None),
("uis.dat", "binomial", 8, 10, None),
]
trial = 0
for i in range(3):
for (csvFilename, family, y, timeoutSecs, x) in csvFilenameList:
csvPathname = "logreg/umass_statdata/" + csvFilename
kwargs = {"n_folds": 2, "response": y, "family": family, "alpha": 1, "lambda": 1e-4}
parseResult = h2i.import_parse(
bucket="smalldata", path=csvPathname, schema="put", timeoutSecs=timeoutSecs
)
if x is not None:
ignored_cols = h2o_cmd.createIgnoredCols(key=parseResult["destination_key"], cols=x, response=y)
kwargs["ignored_cols"] = ignored_cols
start = time.time()
glm = h2o_cmd.runGLM(parseResult=parseResult, timeoutSecs=timeoutSecs, **kwargs)
h2o_glm.simpleCheckGLM(self, glm, None, **kwargs)
print "glm end (w/check) on ", csvPathname, "took", time.time() - start, "seconds"
trial += 1
print "\nTrial #", trial
