End-to-End Machine Learning using AWS SageMaker

Overview

• Step 1: Introduction to AWS SageMaker

  • Overview of SageMaker, S3, EC2 & IAM features and capabilities
  • Setting up AWS environment and SageMaker instance

• Step 2: GitHub Setup & Data Cleaning

  • Setting up local & remote repository using GitHub
  • Data Cleaning using Numpy and Pandas best practices

• Step 3: Exploratory Data Analysis

  • Understanding the workflow of systematically analyzing datasets
  • Understanding the various plots, statistical measures and hypothesis tests to analyze datasets
  • Exploring a custom EDA module for convenience and significantly reduce complexity of analyzing datasets
  • Performing in-depth analysis of various kinds of numeric, categorical and date-time variables
  • Leveraging statistical measures, hypothesis tests, and univariate, bivariate and multivariate plots

• Step 4: Feature Engineering and Data Preprocessing

  • Understanding feature engineering teachniques for different types of variables
  • Creating scikit-learn compatible custom classes and functions
  • Using advanced scikit-learn features for feature engineering and data preprocessing such as:
    • Pipeline
    • Feature Union
    • Function Transformer
    • Column Transformer

• Step 5: Model Training and Deployment

  • Training and Tuning a machine learning model on SageMaker
  • Using S3 buckets for storage and EC2 for computing purposes
  • Creating a web application from scratch and deploying over cloud using Streamlit