97 Things About Ethics Everyone in Data Science Should Know

Chapter 32. Limit the Viewing of Customer Information by Use Case and Result Sets

Robert J. Abate

VP & CDO, Global IDs Inc.

One of the challenges of data science today is the availability of many datasets that can be assembled together for a use case (i.e., to provide a 360-degree view of the customer), such that the resulting integrated dataset creates toxic combinations of data, including information that might be misused if in the wrong hands.

Consider that if you combine customer information and shopping history (in the case of a retailer of consumer product goods with direct-to-customer sales) with US Census Bureau information and CDC natality (birth rate) statistics, you can determine a lot about a household—too much, in fact. For example, you can determine the members of the household and their education levels and incomes, the items they’ve purchased on a regular basis, the age of children in the household, and so on.

We have learned that the right people should see the right data at the right time with the right quality (a standard in data governance programs), but has your organization considered the use cases for information at this level of detail? One way to limit people from viewing Personally Identifiable Information (PII), or to limit the creation of toxic combinations of information (e.g., name, address, age, and phone number), is to “sign data out” from your ...