Prescriptive Analytics

What Is Prescriptive Analytics?

Prescriptive analytics answers the question “What should/can be done?” by using machine learning (ML), modeling, simulation, heuristics, and other methods to predict outcomes and provide decision options. Building upon descriptive and predictive analytics, prescriptive analytics not only provides forecasting and predictions about future events, but what could make them happen. Using this information, analysts can test the impact of strategic decisions to optimize their decision-making processes.

Why Is Prescriptive Analytics Important?

Building on the work of descriptive and predictive analytics, prescriptive analytics can benefit a business by helping them:

• Make informed, fact-based decisions by using real-time and forecasted data
• Understand the likelihood of certain outcomes the impact of business decisions on those outcomes, and use it to plan what to do and how
• Save resources and boost efficiency by allowing AI to curate and process data into actionable scenarios
• Create reproducible and scalable processes from prescriptive analytics to make decisions using near-time data
• Answer the most complex business questions such as demand forecasting, risk assessment, and what-if scenarios

How Prescriptive Analytics Works

Prescriptive analytics is the final step in business analytics and leverages the outcomes of several statistical methods and the power of AI. While descriptive analytics answers, “What has happened?” and predictive analytics answers, “What could happen?” prescriptive analytics answers, “What should we do?” and “How will our decisions affect future performance?” It gives analysts and decision makers the power
to positively and confidently impact future outcomes through optimization models and iterative machine learning.

Prescriptive analytics can benefit any data-driven business and is highly utilized in fields where data is constantly changing and decisions can have wide-ranging impact.

• In healthcare, prescriptive analytics can help in both administration and patient care. A pharmaceutical company can use prescriptive analytics to reduce the costs of testing by finding the best subjects for a clinical trial, while a hospital might use it to provide attention to patients who need it most by seeing who has the highest risk of re-admission.
• In transportation, an airline can automatically adjust pricing and availability based on various factors including weather, demand, and oil prices.
• In publishing, an outlet can decide what to publish and if a piece will be popular based on search and social data for similar topics.
• In human resources, online training can be adjusted in real time based on an employee’s performance on each lesson.

Getting Started With Prescriptive Analytics

Alteryx ML platform provides automated machine learning (AutoML) and feature engineering for prescriptive analytics, giving users the ability to test ML models in a fully guided user experience, without the need for coding complex models. Alteryx ML allows users to:

• Uncover hidden relationships within your prescriptive analytics data with Automated Insight Generation
• Use algorithms like xgBoost, LightGBM, and ElasticNet to uncover features in your data that have the highest impact on model performance
• Create understandable and explainable models and dashboards that can communicate feature importance, impact analysis, and simulation exploration
• Quickly create trusted prescriptive analytics models using pre-defined feature libraries
• Integrate prescriptive analytics models into business processes through Alteryx’s end-to-end business platform