Exploring AF Analytics for Advanced Analysis and Prediction

In this lab, we’ll complete three exercises demonstrating how AF Analytics can serve as an advanced analytics workbench

About this course

AF Analytics offers distinctive advantages when preparing time-series data for advanced analytics.  It can play an essential role in testing, evaluating and operationalizing developed models.  In addition, the openness of the PI System supports several different data access methods, meeting the needs of data engineers or scientists.

In this hands-on lab, we’ll complete three exercises demonstrating how AF Analytics can serve as an advanced analytics workbench.  For comparison, we will employ different data access methods, PI Integrator for Business Analytics and PI SQL Client.  (Exercises are replicated in the Appendix using PI OLEDB Enterprise and the PI Web API.)  Predictive models using PI Data will be developed in Python, a readily available open source application that is widely used for statistical modelling and analysis.

By clicking 'register' you will be able to view the content of the lab. If you want to go through the materials in a pre-configured Training Cloud Environment, you can purchase the Training Cloud Environment Subscription (Monthly or Yearly). This subscription will grant you access to over 50 similar lab environments with associated workbooks. For more information, click the links below:

The lab consists of three exercises.  The same statistical model is developed in each exercise, however different features of the PI Infrastructure are used in each case:

Exercise 1 - Develop a Predictive Model for a Single Transformer

  • Data preparation - PI Integrator for Business Analytics (BA)
  • Model testing - AF Analytics “Preview Results” – No PI Tags required
  • Model evaluation – Microsoft Power BI

Exercise 2 - Develop a Predictive Model for Multiple Transformers

  • Data preparation - PI SQL Client
  • Model testing – AF Analytics Backfill – PI Tags required
  • Model evaluation - PI Vision Collections

Exercise 3 - Operationalize Forecast Model for Multiple Transformers

  • Operationalization – AF Analytics, PI Future Data and PI Vision

Additional data access methods and learning resources are described in the appendix.

  • Appendix A - Example 2 using PI OLEDB Enterprise
  • Appendix B - Example 2 using PI Web API
  • Appendix C – Learning Resources

Download Workbook.

Course Outline

  • Getting Started
  • Learning Objectives
  • Lab Setup
  • Learn
  • Workbook - Exploring AF Analytics for Advanced Analysis and Prediction
  • Access to Training Cloud Environment

About this course

AF Analytics offers distinctive advantages when preparing time-series data for advanced analytics.  It can play an essential role in testing, evaluating and operationalizing developed models.  In addition, the openness of the PI System supports several different data access methods, meeting the needs of data engineers or scientists.

In this hands-on lab, we’ll complete three exercises demonstrating how AF Analytics can serve as an advanced analytics workbench.  For comparison, we will employ different data access methods, PI Integrator for Business Analytics and PI SQL Client.  (Exercises are replicated in the Appendix using PI OLEDB Enterprise and the PI Web API.)  Predictive models using PI Data will be developed in Python, a readily available open source application that is widely used for statistical modelling and analysis.

By clicking 'register' you will be able to view the content of the lab. If you want to go through the materials in a pre-configured Training Cloud Environment, you can purchase the Training Cloud Environment Subscription (Monthly or Yearly). This subscription will grant you access to over 50 similar lab environments with associated workbooks. For more information, click the links below:

The lab consists of three exercises.  The same statistical model is developed in each exercise, however different features of the PI Infrastructure are used in each case:

Exercise 1 - Develop a Predictive Model for a Single Transformer

  • Data preparation - PI Integrator for Business Analytics (BA)
  • Model testing - AF Analytics “Preview Results” – No PI Tags required
  • Model evaluation – Microsoft Power BI

Exercise 2 - Develop a Predictive Model for Multiple Transformers

  • Data preparation - PI SQL Client
  • Model testing – AF Analytics Backfill – PI Tags required
  • Model evaluation - PI Vision Collections

Exercise 3 - Operationalize Forecast Model for Multiple Transformers

  • Operationalization – AF Analytics, PI Future Data and PI Vision

Additional data access methods and learning resources are described in the appendix.

  • Appendix A - Example 2 using PI OLEDB Enterprise
  • Appendix B - Example 2 using PI Web API
  • Appendix C – Learning Resources

Download Workbook.

Course Outline

  • Getting Started
  • Learning Objectives
  • Lab Setup
  • Learn
  • Workbook - Exploring AF Analytics for Advanced Analysis and Prediction
  • Access to Training Cloud Environment