Part 3 Serverless machine learning pipeline

 

A machine learning system is more than just a model and a data set. In this part, you will walk through the steps of engineering an entire machine learning pipeline, starting from the steps involved in automation of feature engineering to hyperparameter optimization and experiment management.

  • In chapter 9, you will explore the use cases around feature selection and feature engineering, learning from case studies to understand the kinds of features that can be created for the DC taxi data set.
  • In chapter 10, you will adopt a PyTorch framework called PyTorch Lightning to minimize the amount of boilerplate engineering code in your implementation. In addition, you will ensure that you can train, validate, and test your PyTorch Lightning-based machine learning model.
  • In chapter 11, you will integrate your machine learning model with the Optuna hyperparameter framework, training alternative models based on Optuna-suggested hyperparameter values and ranking the models according to their loss and metric performance.
  • In chapter 12, you will package your machine learning model implementation into a Docker container in order to run the container through the various stages of the entire machine learning pipeline, starting from the development data set all the way to a trained model ready for production deployment.