predict.py

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn import datasets
import json
import boto3

global app_name
global region
app_name = ''
region = ''

def check_status(app_name):
    sage_client = boto3.client('sagemaker', region_name=region)
    endpoint_description = sage_client.describe_endpoint(EndpointName=app_name)
    endpoint_status = endpoint_description['EndpointStatus']
    return endpoint_status

def query_endpoint(app_name, input_json):
    client = boto3.session.Session().client('sagemaker-runtime', region)

    response = client.invoke_endpoint(
        EndpointName = app_name,
        Body = input_json,
        ContentType = 'application/json; format=pandas-split',
        )

    preds = response['Body'].read().decode('ascii')
    preds = json.loads(preds)
    print('Received response: {}'.format(preds))
    return preds

# Check endpoint status
print('Application status is {}'.format(check_status(app_name)))

# Prepare to give for predictions
iris = datasets.load_iris()
X = iris.data[:, 2:]
y = iris.target
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=101)

# Create test data and make inference from endpoint
query_input = pd.DataFrame(X_train).iloc[[15]].to_json(orient='split')
print(query_input)
predictions = query_endpoint(app_name=app_name, input_json=query_input)