1.1 Introduction: Types of Analytics
The section above notes that EDA is a form of descriptive analytics. Analytical tasks can broadly be categorized in one of three ways: descriptive, predictive, and prescriptive.
Descriptive analytics answers the question “What happened?”
Descriptive analytics tasks could include things like determining the average daily merchandise revenue at Lobster Land for a particular year, the gross revenue for the park throughout the entire season, or the correlation between ride usage and precipitation. Descriptive analytics can range from the very simple (a mean or a median to describe the central tendency of a single variable) to complex statistics that capture the interrelationships among variables.
Predictive analytics answers the question, “What will happen?”
Predictive analytics is often performed with the help of models. A company is using predictive analytics when it builds a regression model to estimate the likely revenue for next quarter, based on its marketing spending this quarter, or when it uses a simple moving average to predict season pass renewals from year to year. Predictive analytics can be useful in a general, macro sense (if Lobster Land can predict the percentage of passholders who will renew, that can help to shape their budgeting decisions), as well as in a specific, micro sense (if Lobster Land predicts that Person A is 95% likely to renew, Person B is 5% likely to renew, and that Person C is 50% likely to renew, they might choose to concentrate on Person C in their one-to-one marketing outreach campaign).
Prescriptive analytics answers the question, “What should happen?”
Prescriptive analytics will not result from algorithms and data tables alone – it requires a mix of quantitative analysis and subjective insight. The particular course of action that a business ought to take at some certain moment is dependent on a complicated mix of factors that could include everything from competitors’ behavior to local government regulations to an understanding of the various personalities involved in a decision.
In many organizations, the analytics team will not even make the final call regarding major business decisions. Nonetheless, it is likely that the decision maker, such as the CEO or COO, will turn to analytics and ask “What do you think? What is your recommendation?” Therefore, it is essential that analytics professionals not only deliver the “what” that comes from their findings, but also the “So what?” factor (Why should this business care? Why is this finding relevant?)