def generate_submodel_gpr_compile_task(self, submodel_id, location_ids,
upstream_tasks):
job_hash_name = "m5q0_{}_submodel_{}_gpr_compile".format(
self.version_id, submodel_id)
slots = 5
mem_free = 10
process_timeout = 10000
runfile = "{}/07_append_gpr.py".format(self.code_dir)
args = [
"--version_id",
str(self.version_id), "--submodel_id",
str(submodel_id), "--location_ids",
' '.join([str(l) for l in location_ids])
]
return PythonOperator(job_hash_name,
slots=slots,
mem_free=mem_free,
project=self.cluster_project,
runfile=runfile,
args=args,
process_timeout=process_timeout,
path_to_python_binary=self.python_binary_path,
upstream_tasks=upstream_tasks)
def generate_submodel_gpr_task(self, submodel_id, location_id, ihme_loc_id,
upstream_tasks):
job_hash_name = "m5q0_{}_submodel_{}_gpr_{}".format(
self.version_id, submodel_id, location_id)
slots = 5
mem_free = 10
process_timeout = 10000
runfile = "{}/06_fit_gpr.py".format(self.code_dir)
args = [
"--version_id",
str(self.version_id), "--submodel_id",
str(submodel_id), "--location_id",
str(location_id), "--ihme_loc_id", ihme_loc_id
]
return PythonOperator(
job_hash_name,
slots=slots,
mem_free=mem_free,
project=self.cluster_project,
runfile=runfile,
args=args,
process_timeout=process_timeout,
path_to_python_binary=self.python_gpr_binary_path,
upstream_tasks=upstream_tasks)
def generate_assess_bias_task(self, upstream_tasks):
job_hash_name = "m5q0_{}_assess_bias".format(self.version_id)
slots = 5
mem_free = 10
process_timeout = 10000
runfile = "{}/02_assess_vr_bias.py".format(self.code_dir)
args = ["--version_id", str(self.version_id)]
return PythonOperator(job_hash_name,
slots=slots,
mem_free=mem_free,
project=self.cluster_project,
runfile=runfile,
args=args,
process_timeout=process_timeout,
path_to_python_binary=self.python_binary_path,
upstream_tasks=upstream_tasks)
def generate_sex_gpr_task(version_id, ihme_loc_id, region_name, upstream_tasks):
job_hash_name = "agesex_{}_sex_gpr_{}".format(
version_id, ihme_loc_id)
slots = 5
mem_free = 10
project = "CLUSTER_PROJECT"
path_to_python_binary="FILEPATH"
runfile = ("FILEPATH"
"05_predict_sex_model_gpr.py")
args = ["--version_id", str(version_id),
"--region_name", region_name,
"--ihme_loc_id", ihme_loc_id]
process_timeout = 10000
return PythonOperator(
job_hash_name, slots=slots, mem_free=mem_free, project=project,
runfile=runfile, args=args, process_timeout=process_timeout,
path_to_python_binary=path_to_python_binary,
upstream_tasks=upstream_tasks)
def generate_save_draws_task(self, location_id, upstream_tasks):
job_hash_name = "m5q0_{}_save_draws_{}".format(self.version_id,
location_id)
slots = 5
mem_free = 10
process_timeout = 10000
runfile = "{}/09_save_draws.py".format(self.code_dir)
args = ["--version_id", self.version_id, "--location_id", location_id]
return PythonOperator(job_hash_name,
slots=slots,
mem_free=mem_free,
project=self.cluster_project,
runfile=runfile,
args=args,
process_timeout=process_timeout,
path_to_python_binary=self.python_binary_path,
upstream_tasks=upstream_tasks)
def generate_u5_task(location_id, ihme_loc_id, version_id, version_age_sex_id,
upstream_tasks):
job_hash_name = "u5_{}_{}".format(version_id, location_id)
slots = 12
mem_free = 24
project = "CLUSTER_PROJECT"
runfile = code_dir + "01_u5.py"
args = [
"--location_id",
str(int(location_id)), "--ihme_loc_id", ihme_loc_id, "--version_id",
str(int(version_id)), "--version_age_sex_id",
str(int(version_age_sex_id))
]
process_timeout = 10000
return PythonOperator(job_hash_name,
slots=slots,
mem_free=mem_free,
project=project,
runfile=runfile,
args=args,
process_timeout=process_timeout,
path_to_python_binary=path_to_python_binary,
upstream_tasks=upstream_tasks)