2 Should you send a purchase order to a technical approver?

 

This chapter covers:

  • Identifying a machine-learning opportunity
  • Identifying what data is required and how much data you’ll need
  • Building a machine-learning system
  • Using the machine-learning system to make decisions

In this chapter you’ll work through an end-to-end machine-learning scenario that makes a decision about whether to send an order to a technical approver or not. And you’ll do it without writing any rules. All you’ll feed into the machine learning system is a dataset consisting of 1,000 historical orders and a flag that indicates whether that order was sent to a technical approver or not. The machine learning system will figure out the patterns from those 1,000 examples and, when given a new order, will be able to correctly determine whether to send the order to a technical approver.

2.1  The Decision

2.2  The data

2.3  Putting on your training wheels

2.4  Run the Jupyter notebook and make predictions

2.5  Part 1: Load and examine the data

2.6  Part 2: Get the data into the right shape

2.7  Part 3: Create training, validation and test data sets

2.8  Part 4: Train the model

2.9  Part 5: Host the model

2.10  Part 6: Test the model

2.11  Delete the endpoint and shut down your notebook instance

2.11.1  Delete the endpoint

2.11.2  Shut down the notebook instance

2.12  Summary