row.update(kwargs)
row.update(glm)
row.update(val)
row.update({'wall_clock_secs': wall_clock_secs})
row.update({'java_heap_GB': java_heap_GB})
csvWrt.writerow(row)
h2o.nodes[0].remove_key(k)
finally:
output.close()
def parse_file(f):
v = h2o.nodes[0].import_files()['succeeded'][0]
return h2o.nodes[0].parse(v['key'],timeoutSecs=3600)['destination_key']
if __name__ == '__main__':
h2o.parse_our_args()
files = None
if is_ec2():
files = ec2_files
h2o_hosts.build_cloud_with_hosts()
else:
files = local_files
h2o_hosts.build_cloud_with_hosts(use_hdfs=True,base_port=54321)
# want to ignore columns with missing values, since GLM throws away those rows, (won't analyze as many rows)
# Distance, CRSEElapsedTime has some...I guess ignore
# column Year 0 type: int
# column Month 1 type: int
# column DayofMonth 2 type: int
# column DayOfWeek 3 type: int
# column DepTime 4 type: int num_missing_values: 2302136
timeoutSecs = 7200,
**kwargs)
h2j.pollWaitJobs(timeoutSecs=4800, pollTimeoutSecs=4800, retryDelaySecs=2)
pcaTime = time.time() - pcaStart
cmd = 'bash startloggers.sh ' + json + ' stop_'
#stop all loggers
os.system(cmd)
row.update({'pcaBuildTime' : pcaTime})
csvWrt.writerow(row)
finally:
output.close()
if __name__ == '__main__':
dat = sys.argv.pop(-1)
debug = sys.argv.pop(-1)
build = sys.argv.pop(-1)
json = sys.argv[-1].split('/')[-1]
fp = 'Airlines' if 'Air' in dat else 'AllBedrooms'
h2o.parse_our_args()
h2o_hosts.build_cloud_with_hosts()
if dat == 'Air1x' : fs = files['Airlines']['train'][0]
if dat == 'Air10x' : fs = files['Airlines']['train'][1]
if dat == 'Air100x' : fs = files['Airlines']['train'][2]
if dat == 'AllB1x' : fs = files['AllBedrooms']['train'][0]
if dat == 'AllB10x' : fs = files['AllBedrooms']['train'][1]
if dat == 'AllB100x' : fs = files['AllBedrooms']['train'][2]
doPCA(fs, fp)
h2o.tear_down_cloud()
