Chapter 2: Create, Train, and Predict a Machine Learning Model

This section covers the steps required create and train a machine learning model. And then use the training data to predict planned transit time for your shipments.

This process includes the following steps:

1. Creating a Machine Learning Project
2. Creating a Machine Learning Scenario
3. Sending Data to Analytics
4. Exporting Data to Machine Learning
5. Performing Training
6. Performing Prediction
7. Implementing Automated Workflows

Creating a Machine Learning Project

First, you create a Logistics Machine Learning project with the following details:

1. Go to Logistics Machine Learning > Machine Learning Project.
2. Click New.
3. Enter a Project ID.
4. Select an Platform of External System.
5. Select an Model Type:
   • Planned ETA Prediction: This model type uses shipment-related attributes as the input to the model at the planning time or when the actual departure time at the source is known. This model type provides a base prediction before any in-transit tracking events are received.
   • 'Planned and Event based ETA Prediction: This model type uses planned ETA prediction (as described above) and adds in-transit tracking event information to further refine the predicted ETA.
6. Specify a Shipment Saved Query. This defines the scope of the historical data that is used for training the model.
7. Specify a Logic Configuration ID. 
   
   The value entered for the Machine Learning External Systems ID paramter is the OTM External System ID that points to IoT rather than Object Storage.
8. Click New Machine Learning Scenario and create a new scenario as specified in the next section.

Creating a Machine Learning Scenario

When you click the New Machine Learning button, you see the machine learning scenario creation page. Since you clicked New Machine Learning Scenario on the project, the scenario is added to the project.

1. Select a Model Type depending on what is on the machine learning project:
   1. If the machine learning project has a Model Type of Planned ETA Prediction, you see a single option of Planned ETA Prediction. This option creates machine learning models that only consider shipment-level histories and are good for predicting shipments that are either not executed or do not have any tracking events.
   2. If the machine learning project has a Model Type of Planned and Event based ETA Prediction, you see two options: Event Based ETA Prediction and Planned ETA Prediction.
      
      Event Based ETA Prediction creates machine learning models that not only considers shipment-level information but also tracking events. These models are good for predicting shipments that are in transit.
7. Create filters.
   • Shipment Filter
   • Outlier Filter
8. Add Excluded Columns.

Sending Data to Analytics

Next, you use the Load Data into Analytics action to send the data to the Logistics Network Modeling subject area in Oracle Transportation Intelligence. 

Once the data is in analytics, you can view data distribution on Analytics dashboards. Validate the data and adjust Scope if needed by modifying Shipment Saved Queries. If you adjust data scope, you must run this action again.

Exporting Data to Machine Learning

Once the data is in analytics, you send it from analytics to the machine learning server using the Export to Machine Learning Services action.

Note: For the External System ID, enter theID that points to Object Storage.

Performing Training

Once the data has been transferred to the machine leaning IoT server, you are ready to run the Perform Training action on 1 or more Machine Learning scenarios. When the training is completed, view the training results to see accuracy and feature important scores.

Performing Prediction

Next, you select shipments and run the Perform Prediction action against your preferred Machine Learning project and scenario.

Implementing Automated Workflows

Finally, you evaluate all of this data and set up process management to automate the above steps.