Data Science Basics and Management Accounting

Learning Objective 1 Explain how management accountants can work with data scientists to create value

• use data to predict outcomes and impact decision making

This chapter covers foundational concepts of data science for predictive modeling to aid decision making. Data science refers to the use of data analytics to draw conclusions from data. Predictive modeling is a data science technique used to make predictions based on past or current data.

Outcome Prediction

The increasing availability of inexpensive data storage and Web-based (cloud) computing power allows data scientists to use very large datasets to train sophisticated algorithmic models. These models learn from training data (thousands or millions of records) and can then predict a new record according to some feature of interest. For example, a feature of interest to a telecommunications provider is churn, that is, whether customers will switch providers in the next quarter. A cellular telecommunications operator, such as AT&T or T-Mobile, can use data, such as number of calls made, number of dropped calls, and number of family members who are on a family plan, to predict which customers are most likely to leave the company in the next 3 months. In this example, the feature churn takes two possible values (binary): likely to leave in 3 months (1), not likely to leave in 3 months (0). The management accountant can then work with marketing to determine the costs and benefits of keeping those customers or alternatively letting them leave. The ability to accurately predict outcomes can directly impact how an organization crafts its strategy.

As shown in Exhibit 11-1, data science sits at the intersection of (1) computer science and data skills, (2) math and statistics, and, critically, (3) substantive expertise in a particular area of interest or domain such as industry and management accounting knowledge. As the churn example in the telecommunications industry indicates, deep knowledge about the economics of a business and industry is important for formulating relevant questions for data science to address. Exhibit 11-1 also makes another point. Management accountants need to understand some of the computer science and statistics tools used in data science so that they can effectively interact with other members of the data science team. The circle representing the expertise needed by management accountants overlaps with computer science and statistical inference skills.