The case for personal data-driven decision making

Data-driven decision making (D3M) has shown great promise in professional pursuits such as business and government. Here, policymakers collect and analyze data to make their operations more efficient and equitable. Progress in bringing the benefits of D3M to everyday life has been slow. For example, a student asks, "If I pursue an undergraduate degree at this university, what are my expected lifetime earnings?". Presently there is no principled way to search for this, be-cause an accurate answer depends on the student and school. Such queries are personalized, winnowing down large datasets for specific circumstances, rather than applying welldefined predicates. They predict decision outcomes by extrapolating from relevant examples. This vision paper introduces a new approach to D3M that is designed to empower the individual to make informed choices. Here, we highlight research opportunities for the data management community arising from this proposal.