Data Hunches: Incorporating Personal Knowledge into Visualizations

TL;DR

This paper introduces the concept of data hunches—analysts' implicit knowledge about datasets—and proposes visualization techniques to record, communicate, and leverage this knowledge for better collaboration and understanding.

Contribution

It defines data hunches, discusses their importance, and provides guidelines and techniques for visualizations to capture and share analysts' implicit dataset knowledge.

Findings

Techniques for recording data hunches in visualizations

Guidelines for designing visualizations supporting data hunches

Potential to improve collaboration and learning among analysts

Abstract

The trouble with data is that it frequently provides only an imperfect representation of a phenomenon of interest. Experts who are familiar with their datasets will often make implicit, mental corrections when analyzing a dataset, or will be cautious not to be over-confident in any findings if caveats are present. However, the implicit knowledge about the caveats of a dataset are typically not collected in a structured way, which is problematic especially when teams work together who might have knowledge about different aspects of a dataset. In this work, we define such analyst's knowledge about datasets as data hunches. We discuss the implications of data hunches and propose a set of techniques for recording and communicating data hunches through data visualization. Furthermore, we provide guidelines for designing visualizations that support recording and visualizing data hunches. We envision that data hunches will empower analysts to externalize their knowledge, facilitate collaboration and communication, and support the ability to learn from others' data hunches.