Skip to content

Update xgboost churn neo example for sagemaker v2 #1591

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 11 commits into from
Oct 13, 2020
Merged

Update xgboost churn neo example for sagemaker v2 #1591

Changes from all commits
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
e70ce78
Update xgboost churn neo example for sagemaker v2
Oct 5, 2020
cb38428
Remove use of latest version
Oct 6, 2020
ddd1507
Merge branch 'master' into fix_xgboost_churn_neo
Oct 7, 2020
2509469
Add sagemaker installation command and remove duplicate import
Oct 7, 2020
0c8c2d4
Use sagemaker pysdk v2
Oct 7, 2020
05ac271
Merge branch 'master' into fix_xgboost_churn_neo
Oct 9, 2020
f02f0ed
Merge branch 'master' into fix_xgboost_churn_neo
Oct 9, 2020
b30a9a2
Add setup and cleanup steps
Oct 10, 2020
c6ed70d
clear output
Oct 10, 2020
99579c8
Merge branch 'master' into fix_xgboost_churn_neo
Oct 13, 2020
e72c9ae
Revert kernel metadata
Oct 13, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
139 changes: 92 additions & 47 deletions sagemaker_neo_compilation_jobs/xgboost_customer_churn/xgboost_customer_churn_neo.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,7 +36,42 @@
"\n",
"## Setup\n",
"\n",
"_This notebook was created and tested on an ml.m4.xlarge notebook instance._\n",
"_This notebook was created and tested on an ml.m4.xlarge notebook instance. It assumes that the 'sagemaker' package is already installed in the environment._ \n",
"\n",
"_It has been tested on the following versions of important python packages:_\n",
"\n",
"_sagemaker>=2.14_,\n",
"_pandas==1.1.2_"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import sys\n",
"import sagemaker\n",
"import importlib\n",
"orig_sm_version = sagemaker.__version__\n",
"\n",
"with open('orig_sm_version.txt', \"w\") as f:\n",
" f.write(orig_sm_version)\n",
"\n",
"%pip install --no-cache-dir -qU \"sagemaker>=2.14.0\""
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Please restart the kernel to use the updated sagemaker package and continue below"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"\n",
"Let's start by specifying:\n",
"\n",
Expand Down Expand Up @@ -86,8 +121,7 @@
"import json\n",
"from IPython.display import display\n",
"from time import strftime, gmtime\n",
"import sagemaker\n",
"from sagemaker.predictor import csv_serializer"
"from sagemaker.serializers import CSVSerializer"
]
},
{
Expand Down Expand Up @@ -325,8 +359,8 @@
"metadata": {},
"outputs": [],
"source": [
"from sagemaker.amazon.amazon_estimator import get_image_uri\n",
"container = get_image_uri(boto3.Session().region_name, 'xgboost')"
"from sagemaker import image_uris\n",
"container = image_uris.retrieve(framework='xgboost', region=boto3.Session().region_name, version='1')"
]
},
{
Expand All @@ -342,8 +376,8 @@
"metadata": {},
"outputs": [],
"source": [
"s3_input_train = sagemaker.s3_input(s3_data='s3://{}/{}/train'.format(bucket, prefix), content_type='csv')\n",
"s3_input_validation = sagemaker.s3_input(s3_data='s3://{}/{}/validation/'.format(bucket, prefix), content_type='csv')"
"s3_input_train = sagemaker.inputs.TrainingInput(s3_data='s3://{}/{}/train'.format(bucket, prefix), content_type='csv')\n",
"s3_input_validation = sagemaker.inputs.TrainingInput(s3_data='s3://{}/{}/validation/'.format(bucket, prefix), content_type='csv')"
]
},
{
Expand All @@ -370,8 +404,8 @@
"\n",
"xgb = sagemaker.estimator.Estimator(container,\n",
" role, \n",
" train_instance_count=1, \n",
" train_instance_type='ml.m4.xlarge',\n",
" instance_count=1, \n",
" instance_type='ml.m4.xlarge',\n",
" output_path='s3://{}/{}/output'.format(bucket, prefix),\n",
" sagemaker_session=sess)\n",
"xgb.set_hyperparameters(max_depth=5,\n",
Expand All @@ -393,7 +427,7 @@
"---\n",
"## Host\n",
"\n",
"Now that we've trained the algorithm, let's create a model and deploy it to a hosted endpoint."
"Now that we've trained the algorithm, let's create a model and deploy it to a hosted endpoint. We'll also need to setup serializers and deserializers for passing our `test_data` NumPy arrays to the model behind the endpoint."
]
},
{
Expand All @@ -403,7 +437,7 @@
"outputs": [],
"source": [
"xgb_predictor = xgb.deploy(initial_instance_count=1,\n",
" instance_type='ml.m4.xlarge')"
" instance_type='ml.m4.xlarge', serializer=CSVSerializer())"
]
},
{
Expand All @@ -412,25 +446,14 @@
"source": [
"### Evaluate\n",
"\n",
"Now that we have a hosted endpoint running, we can make real-time predictions from our model very easily, simply by making an http POST request. But first, we'll need to setup serializers and deserializers for passing our `test_data` NumPy arrays to the model behind the endpoint."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"xgb_predictor.content_type = 'text/csv'\n",
"xgb_predictor.serializer = csv_serializer\n",
"xgb_predictor.deserializer = None"
"Now that we have a hosted endpoint running, we can make real-time predictions from our model very easily, simply by making an http POST request."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now, we'll use a simple function to:\n",
"We'll use a simple function to:\n",
"1. Loop over our test dataset\n",
"1. Split it into mini-batches of rows \n",
"1. Convert those mini-batchs to CSV string payloads\n",
Expand All @@ -449,10 +472,10 @@
" predictions = ''\n",
" for array in split_array:\n",
" predictions = ','.join([predictions, xgb_predictor.predict(array).decode('utf-8')])\n",
"\n",
" \n",
" return np.fromstring(predictions[1:], sep=',')\n",
"\n",
"predictions = predict(test_data.as_matrix()[:, 1:])"
"predictions = predict(test_data.values[:, 1:])"
]
},
{
Expand Down Expand Up @@ -604,6 +627,15 @@
"**Important. If the following command result in a permission error, scroll up and locate the value of execution role returned by `get_execution_role()`. The role must have access to the S3 bucket specified in ``output_path``.**"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"'/'.join(xgb.output_path.split('/')[:-1])"
]
},
{
"cell_type": "code",
"execution_count": null,
Expand All @@ -615,7 +647,7 @@
" input_shape={'data':[1, 69]},\n",
" role=role,\n",
" framework='xgboost',\n",
" framework_version='0.7',\n",
" framework_version='0.9',\n",
" output_path=output_path)"
]
},
Expand All @@ -627,7 +659,9 @@
"\n",
"We can deploy this compiled model, note that we need to use the same instance that the target we used for compilation. This creates a SageMaker endpoint that we can use to perform inference. \n",
"\n",
"The arguments to the ``deploy`` function allow us to set the number and type of instances that will be used for the Endpoint. Make sure to choose an instance for which you have compiled your model, so in our case `ml_c5`. Neo API uses a special runtime (DLR runtime), in which our optimzed model will run."
"The arguments to the ``deploy`` function allow us to set the number and type of instances that will be used for the Endpoint. Make sure to choose an instance for which you have compiled your model, so in our case `ml_c5`. Neo API uses a special runtime (DLR runtime), in which our optimzed model will run.\n",
"\n",
"The compiled model accepts CSV content type:"
]
},
{
Expand All @@ -638,18 +672,15 @@
"source": [
"# known issue: need to manually specify endpoint name\n",
"compiled_model.name = 'deployed-xgboost-customer-churn'\n",
"# There is a known issue where SageMaker SDK locates the incorrect docker image URI for XGBoost\n",
"# For now, we manually set Image URI\n",
"compiled_model.image = get_image_uri(sess.boto_region_name, 'xgboost-neo', repo_version='latest')\n",
"compiled_predictor = compiled_model.deploy(initial_instance_count = 1, instance_type = 'ml.c5.4xlarge')"
"\n",
"compiled_predictor = compiled_model.deploy(initial_instance_count = 1, instance_type = 'ml.c5.4xlarge', serializer=CSVSerializer())"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Making an inference request\n",
"The compiled model accepts CSV content type:"
"### Making an inference request\n"
]
},
{
Expand All @@ -658,9 +689,22 @@
"metadata": {},
"outputs": [],
"source": [
"compiled_predictor.content_type = 'text/csv'\n",
"compiled_predictor.serializer = csv_serializer\n",
"compiled_predictor.deserializer = None"
"def optimized_predict(data, rows=500):\n",
" split_array = np.array_split(data, int(data.shape[0] / float(rows) + 1))\n",
" predictions = ''\n",
" for array in split_array:\n",
" predictions = ','.join([predictions, compiled_predictor.predict(array).decode('utf-8')])\n",
"\n",
" return np.fromstring(predictions[1:], sep=',')\n",
"\n",
"predictions = optimized_predict(test_data.values[:, 1:])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Clean up sagemaker version"
]
},
{
Expand All @@ -669,15 +713,16 @@
"metadata": {},
"outputs": [],
"source": [
"def optimized_predict(data, rows=500):\n",
" split_array = np.array_split(data, int(data.shape[0] / float(rows) + 1))\n",
" predictions = ''\n",
" for array in split_array:\n",
" predictions = ','.join([predictions, compiled_predictor.predict(array).decode('utf-8')])\n",
"\n",
" return np.fromstring(predictions[1:], sep=',')\n",
"# rollback the SageMaker Python SDK to the kernel's original version\n",
"with open('orig_sm_version.txt', \"r\") as f:\n",
" orig_sm_version = f.read()\n",
"\n",
"predictions = optimized_predict(test_data.as_matrix()[:, 1:])"
"print(\"Original version: {}\".format(orig_sm_version))\n",
"print(\"Current version: {}\".format(sagemaker.__version__))\n",
"s = 'sagemaker=={}'.format(orig_sm_version)\n",
"print(\"Rolling back to... {}\".format(s))\n",
"%pip install --no-cache-dir -qU {s}\n",
"%rm orig_sm_version.txt"
]
},
{
Expand All @@ -695,8 +740,8 @@
"metadata": {},
"outputs": [],
"source": [
"sagemaker.Session().delete_endpoint(xgb_predictor.endpoint)\n",
"sagemaker.Session().delete_endpoint(compiled_predictor.endpoint)"
"xgb_predictor.delete_endpoint()\n",
"compiled_predictor.delete_endpoint()"
]
}
],
Expand Down