Created
June 7, 2022 05:18
-
-
Save randyphoa/289527fe93d83469257cb771f80655c3 to your computer and use it in GitHub Desktop.
Deploy R in online mode - deploy python function to call R
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
params = { | |
"wml_credentials": wml_credentials, | |
"space_id": SPACE_ID, | |
"script": "predict.R", | |
"required_assets": ["iris_xgb.model"], | |
"r_packages": ["jsonlite", "xgboost"], | |
} | |
def r_function(params=params): | |
""" | |
Deployable Python function which downloads the required assets from a WML deployment space and runs the specified R Script. | |
""" | |
import json | |
import subprocess | |
from ibm_watson_machine_learning import APIClient | |
wml_client = APIClient(params["wml_credentials"]) | |
wml_client.set.default_space(params["space_id"]) | |
script = params["script"] | |
assets = {x["metadata"]["name"]: x["metadata"]["asset_id"] for x in wml_client.data_assets.get_details()["resources"]} | |
# download required assets | |
wml_client.data_assets.download(assets[script], script) | |
for x in params["required_assets"]: | |
wml_client.data_assets.download(assets[x], x) | |
# install R and other libraries | |
# subprocess.run(["mamba", "install", "-c", "conda-forge", "r-base", "r-essentials"]) | |
subprocess.run(["Rscript", "-e", 'install.packages(c("jsonlite", "data.table", "stringi", "stringr"), repos="https://cloud.r-project.org")']) | |
if "xgboost" in params["r_packages"]: | |
params["r_packages"].remove("xgboost") | |
subprocess.run(["Rscript", "-e", 'install.packages("https://cloud.r-project.org/src/contrib/Archive/xgboost/xgboost_1.5.2.1.tar.gz", repos=NULL, type="source")']) | |
subprocess.run(["Rscript", "-e", 'install.packages(c(' + ",".join([f'"{x}"' for x in params["r_packages"]]) + '), repos="https://cloud.r-project.org")']) | |
def score(payload): | |
d = payload["input_data"][0]["values"][0] | |
inputs = d["inputs"] | |
inputs = json.dumps(inputs) | |
result = subprocess.run(["Rscript", "--vanilla", script, inputs], stdout=subprocess.PIPE, stderr=subprocess.PIPE) | |
out = result.stdout.decode("utf-8") | |
try: | |
out = json.loads(out) | |
except: | |
out = result.stdout.decode("utf-8") | |
err = result.stderr.decode("utf-8") | |
return {"predictions": [{"out": out, "err": err}]} | |
return score | |
meta_props = { | |
wml_client.repository.FunctionMetaNames.NAME: "iris xgb", | |
wml_client.repository.FunctionMetaNames.SOFTWARE_SPEC_ID: wml_client.software_specifications.get_uid_by_name("custom-r"), | |
} | |
function_details = wml_client.repository.store_function(function=r_function, meta_props=meta_props) | |
function_uid = wml_client.repository.get_function_id(function_details) | |
meta_props = { | |
wml_client.deployments.ConfigurationMetaNames.NAME: "iris xgb deployment", | |
wml_client.deployments.ConfigurationMetaNames.ONLINE: {}, | |
} | |
wml_client.deployments.create(function_uid, meta_props=meta_props) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment