def test_tc(self):
d = data.fromfile(testfile("testdata5.txt"))
d.discretize()
tasks = [greedy.GreedyLearner(d) for x in range(5)]
tc = ipy1.IPython1Controller("127.0.0.1:10113")
results = tc.run(tasks)
results = result.merge(results)
assert isinstance(results, result.LearnerResult)
def setUp(self):
dat = data.fromfile(testfile("testdata5.txt"))
dat.discretize()
g = greedy.GreedyLearner(dat, max_iterations=100)
g.run()
self.result = g.result
self.tempdir = tempfile.mkdtemp()
self.result.tohtml(self.tempdir)
def test_pebl(numvars, numsamples, greedy_iters, cachesize):
print "Testing with #vars=%d, #samples=%d, iters=%d, cachesize=%d" % (
numvars, numsamples, greedy_iters, cachesize)
config.set('localscore_cache.maxsize', cachesize)
d = data.Dataset(N.random.rand(numsamples, numvars))
d.discretize()
g = greedy.GreedyLearner(d, max_iterations=greedy_iters)
g.run()
return g
def test_example1(self):
outdir = os.path.join(self.tmpdir, "example1-result")
dataset = data.fromfile(testfile("pebl-tutorial-data1.txt"))
dataset.discretize()
learner = greedy.GreedyLearner(dataset)
ex1result = learner.run()
ex1result.tohtml(outdir)
assert os.path.exists(os.path.join(outdir, 'index.html'))
from pebl import data
from pebl import prior
from pebl.learner import greedy
dataset = data.fromfile("output.txt")
dataset.discretize(numbins=3)
node_src = [32 for ii in range(32)] + [3 for ii in range(32)] + [
30 for ii in range(32)
] + [31 for ii in range(32)]
node_dest = range(32) + range(32) + range(32) + range(32)
prior = prior.Prior(33, prohibited_edges=zip(node_src, node_dest))
learner = greedy.GreedyLearner(dataset, prior, max_iterations=120000)
result = learner.run()
result.tohtml()
#!/usr/bin/env python
# Bayesian with five greedy and 5 simulated annearling learners serially
# http://pythonhosted.org/pebl/tutorial.html
from pebl import data, result
from pebl.learner import greedy, simanneal
from pebl.taskcontroller import multiprocess
dataset = data.fromfile("pebl-tutorial-data2.txt")
learners = [ greedy.GreedyLearner(dataset, max_iterations=1000000) for i in range(5) ] + \
[ simanneal.SimulatedAnnealingLearner(dataset) for i in range(5) ]
tc = multiprocess.MultiProcessController(poolsize=2)
results = tc.run(learners)
merged_result = result.merge(results)
merged_result.tofile("example4-result")
from pebl.learner import greedy
from pebl.taskcontroller import ec2
from pebl.test import testfile
help = """Test the EC2 TaskController.
USAGE: test_ec2.py configfile
You need to provide the configfile for use with EC2Controller.
###############################################################################
WARNING for pebl devs:
Do NOT put your configfile under svn.
It contains sensitve information.
###############################################################################
"""
if len(sys.argv) < 2:
print help
sys.exit(1)
d = data.fromfile(testfile("testdata5.txt"))
d.discretize()
tc = ec2.EC2Controller(config=sys.argv[1], min_count=3)
results = tc.run([greedy.GreedyLearner(d, max_time=10) for i in xrange(10)])
results = result.merge(results)
print results
print[r.host for r in results.runs]
start_time = time.time()
classifyUsingNB(trainList, classVariable, testList, testClassVariable)
elapsed_time = time.time() - start_time
print("Time elapsed Gaussian Naive Bayes : ", elapsed_time)
#Decision tree
start_time = time.time()
classifyUsingDecisionTree(trainList, classVariable, testList,
testClassVariable)
elapsed_time = time.time() - start_time
print("Time elapsed Decision : ", elapsed_time)
#logistic regression
solver = 'sag'
multiclass = 'ovr'
start_time = time.time()
classifyUsingLogisticRegression(trainList, classVariable, testList,
testClassVariable, solver, multiclass)
elapsed_time = time.time() - start_time
print("Time elapsed Logistic Regression : ", elapsed_time)
#Bayesian Network using PEBL for the data
start_time = time.time()
dataset = data.fromfile(trainFileNameForBayesian)
dataset.discretize()
learner = greedy.GreedyLearner(dataset)
ex1result = learner.run()
ex1result.tohtml("dataset")
elapsed_time = time.time() - start_time
print("Time elapsed Bayesian Network : ", elapsed_time)
def setUp(self):
config.set('evaluator.missingdata_evaluator', 'exact')
self.data = data.fromfile(
testfile('testdata13.txt')).subset(samples=range(5))
self.learner = greedy.GreedyLearner(self.data, max_iterations=10)
import datasets
DEFAULT_DATASET = datasets.load("greedytest")
DEFAULT_REPORT_DIR = "./report/"
ds = DEFAULT_DATASET
report_dir = DEFAULT_REPORT_DIR
if len(sys.argv) > 1:
ds = datasets.load(sys.argv[1])
if ds == None:
raise Exception("Could not find dataset '%s'" % dataset)
if len(sys.argv) > 2:
report_dir = sys.argv[2]
dataset = ds.dataset
prior = ds.prior
greedy_lrn = greedy.GreedyLearner(dataset, prior, max_iterations=1000)
#anneal_lrn = simanneal.SimulatedAnnealingLearner(dataset, prior)
#results = result.merge(*[ greedy_lrn.run(), anneal_lrn.run() ])
results = greedy_lrn.run()
try:
shutil.rmtree(report_dir)
except:
pass
results.tohtml(report_dir)
def setUp(self):
d = data.fromfile(testfile("testdata5.txt"))
d.discretize()
self.tc = self.tctype(*self.args)
self.tasks = [greedy.GreedyLearner(d, max_iterations=100) for i in xrange(6)]
#!/usr/bin/env python
# http://pythonhosted.org/pebl/tutorial.html
from pebl import data, result
from pebl.learner import greedy
dataset = data.fromfile("pebl-tutorial-data2.txt")
learner1 = greedy.GreedyLearner(dataset, max_iterations=1000000)
learner2 = greedy.GreedyLearner(dataset, max_time=120) # in seconds
result1 = learner1.run()
result2 = learner2.run()
merged_result = result.merge(result1, result2)
merged_result.tofile("example2-result")