Purpose

This tutorial demonstrates how to easily establish the architectural stack for doing data science and machine learning projects on AWS, then deploy the machine learning model for real-time prediction. Primarily, it will cover using Sagemaker to train, evaluate, and deploy deep learning models developed in MXNet. This tutorial also includes the old content, which covers training and evaluating in Deep Learning AMI, then deploying in both Lambda and Batch.

Prerequisites

• Basic knowledge of cloud computing basics
• Basic experience with Python
• Basic knowledge of Data Scient methodologies, specifically CRISP-DM

Recommended Sequence - Sagemaker

1. Start with part0_instructions.md, found in folder 'sagemaker'

Recommended Sequence - Pre-Sagemaker

1. CRISP-DM Parts 3-5, found in folder 'pre-sagemaker/crispdm345-training'
2. CRISP-DM 6 on AWS Lambda, found in folder 'pre-sagemaker/crispdm6-pred-lambda'
3. CRISP-DM 6 on AWS Batch, found in folder 'pre-sagemaker/crispdm6-pred-batch'

Details

Author: Jake Chen ([email protected])