chapter nine

9 Case studies

 

This chapter covers:

  • Review of the topics covered in the book
  • Two case studies featuring companies using machine learning to improve their business
    • Case study 1: Implementing a single machine learning project in your company
    • Case study 2: Implementing machine learning at heart of everything your company does

Throughout the book you have used AWS SageMaker to build solutions to common business problems. The solutions have covered a broad range of scenarios and approaches ranging from the following:

  • Using XGBoost supervised learning to solve an approval routing challenge
  • Reformatting data so that you could use XGBoost again, but this time to predict customer churn
  • Using BlazingText and Natural Language Processing (NLP) to identify whether a tweet should be escalated to your support team
  • Using unsupervised Random Cut Forest to decide whether to query a supplier’s invoice
  • Using DeepAR to predict power consumption based on historical trends
  • Adding datasets such as weather forecasts and scheduled holidays to improve DeepAR’s predictions

In the previous chapter, you learned how to serve your predictions and decisions over the web using AWS’s serverless technology. Now, we’ll wrap it all up with a look at how different companies are implementing machine learning in their business.

9.1  Case Study 1: WorkPac

9.1.1  Designing the project

9.1.2  Stage 1: Preparing and testing the model

9.1.3  Stage 2: Implementing proof of concept

9.1.4  Stage 3: Embedding the process into the company’s operations

9.1.5  Next steps

9.1.6  Lessons learned

9.2  Faethm.ai

9.2.1  AI at the core

9.2.2  Using machine learning to improve processes at Faethm

9.2.3  Stage 1: Get the data

9.2.4  Stage 2: Identify the features

9.2.5  Stage 3: Validate the results

9.2.6  Stage 4: Implement in production

9.3  Conclusion

9.4  Summary