Where descriptive analytics reveals what has happened in the past, prescriptive analytics delivers insight into optimizing future decisions. As data-driven organizations mature, they will begin to apply prescriptive analytics. This area of analytics is often an untapped blue ocean of opportunity.

Predictions, Decisions, and Effects

Prescriptive analytics can help you make decisions, evaluate if/then scenarios, or simply gain a better understanding of a range of possible outcomes. Another common use case for prescriptive analytics is experimentation where actual trial and error would be prohibitively expensive, risky, or time intensive.

Primary applications of prescriptive analytics answer two types of optimization questions.

  • Optimization: How can we achieve the best outcome?
  • Stochastic Optimization: How can we achieve the best outcome and address uncertainty?

Just like its predictive analytics predecessor, prescriptive analytics is ideal for problem solving use cases where there are too many variables, constraints, and data points for the human mind to efficiently evaluate. Optimization and Monte Carlo simulation combines historical data, business rules, mathematical models, variables, constraints, and machine learning algorithms. Numerous scenarios can be evaluated with known and randomized variables.

For example, airline ticket pricing uses prescriptive analytics to sort through complex mixtures of demand curves and purchase timing to present seat prices that will optimize profits but also not deter sales. UPS’s package delivery route optimization is another common use case. There are many other types of problems in operations management, finance, human resources, project management, sales, and marketing that can be solved with optimization models.

Popular examples of prescriptive analytics in action include:

  • Pricing
  • Inventory management
  • Operational resource allocation
  • Production planning
  • Supply chain optimization
  • Transportation and distribution planning
  • Utility management
  • Sales lead assignment
  • Marketing mix optimization
  • Financial planning

How To Get Started

You can start exploring prescriptive analytics with simple what-if scenario analysis. Six years ago I wrote an article on how to do that in Excel using 1) Scenarios, 2) Data Tables, and 3) Goal Seek 4) Excel’s Basic Solver, an add-in to Excel that allows for more variables. Today I’d like to share an example using DataRobot’s what-if analysis extension for Tableau.

The DataRobot What-If Extension empowers you to analyze the cause-and-effect of different variables on a predicted outcome. You can easily compare multiple scenarios with respect to a target outcome to make predictions actionable. To learn more about this Tableau extension, please read my article on DataRobot’s website. If you are attending any of my upcoming events, I’d be happy to show it to you.

DataRobot What If

DataRobot What If

Here is another example of implementing what-if analysis in Qlik Sense. To learn more about this option, please read my article How To Democratize AI in Popular Analytics Tools Part 1. I also showcased these examples and more in the related on demand webinar.

Qlik What If Analysis

To delve deeper in optimization and simulation modeling, check out Modeling and Decision Analysis by Cliff Ragsdale and Management Science: The Art of Modeling with Spreadsheets by Powell and Baker. Those are two of my personal favorite books on these topics. They walk through solving common business problems with model designs and Excel templates.

The optimization design process begins with establishing an appropriate description of the business system to be modeled – setting the objective, control factors, and constraints. Often the model will evolve from an initial conceptual mental model to a visual model diagram, logical model or mathematical model. The image below illustrates an optimization model tutorial designed by Frontline Systems for advertisement planning.

Optimization Model

Source: Frontline Systems

Much like the predictive modeling process, optimization modeling requires a complete and accurate definition of the problem to solve and the stakeholders. You’ll want to work closely with the business to understand who will be using your model, when, and how. After your optimization model is developed, you can use it in an ad-hoc manner or embed it into applications for automated decision making.

That concludes my introduction of prescriptive analytics. For more information, please review the following resources.