Visualized Self

Understanding Self-Reflection: How People Reflect on Personal Data Through Visual Data Exploration

Authors: Eun Kyoung Choe ¹, Bongshin Lee², Haining Zhu¹, Nathalie Henry Riche², Dominikus Baur³Affiliation: ¹Penn State University, ²Microsoft Research, ³Independent ResearcherYear: 2017

My Role: Qualitative UX Researcher

PDF | Presentation

Summary of the Study

Background: Rapid advancements in consumer technologies enable people to collect a wide range of personal data, and visual data exploration is a powerful way to help people reveal meaningful insights about themselves and to facilitate self-reflection. However, most self-tracking tools lack support for self-reflection beyond providing simple feedback.

Goal: Our overarching aim is to support self-trackers in reflecting on their data and gaining rich insights through visual data exploration. As first step towards this goal, we want to understand how do people reflect on their self-tracking data, and what insights do people gain from visual data exploration when they interacting with their personal data using overview and timeline visualizations.

Method: We therefore built a web-based application called Visualized Self, which supports people to integrate personal data from multiple personal informatics systems, explore the data with timeline visualizations, and perform temporal comparisons. Then conducted an in-lab think-aloud study (N = 11) to examine how people reflect on their personal data and what types of insights they gain throughout the reflection.

Deliverable: We discuss lessons learned from studying with Visualized Self, and suggest directions for designing visual data exploration tools for fostering self-reflection.

Data Exploration

With Visualized Self, people can explore time series data on the Trend page, and location data on the Places page. Selecting any of the active summary boxes in (2) leads to the appropriate page based on the data type. Once people select a data source in the Trend page (3), they can see more details about the data, such as tracking duration (4). People can select a particular time frame (5), automatically updating the timeline visualization (6).

We hypothesized that comparing different data sources would lead participants to new in-sights. Thus, with Visualized Self, people can easily compare multiple data streams of the same data type. For example, the data summary (4) and line chart (6) show that steps from Fitbit (black line) is on average ~2,000 higher than steps from Moves (green line). For location data, such as hikes or bike tours, Visualized Self provides a list of paths people have traveled. Once people select a specific path, Visualized Self draws the actual path on a map, helping people recall their activities.

Data Import from Multiple Data Services

We designed Visualized Self to allow for easy data import. Currently, people can import data from Microsoft Band, Fitbit, Aria, RescueTime, RunKeeper, and Moves by a single click (1). Each service has multiple data streams; for example, Fitbit collects step, sleep, and heart rate, among others. For each data service, Visualized Self provides a data summary of each data stream (2) such as tracking duration and range of data values (min, max, average), which was a personal insight type commonly shared during the Quantified Self (QS) presentation.

Contextual Data Management and Data Comparison

Visualized Self provides “weekends” as a default context (7), which can be overlaid on the timeline visualization (shown as gray background). For example, five peaks shown in (6) correspond to weekend hikes. In addition, it allows for the manual addition of contextual information (e.g., vacation, gym) that surfaces during the exploration.

These contexts can also be used as a “factor” to segment data for comparisons on the Comparison page (8). On this page, people can compare data from two different time segments (e.g., this month vs. previous month); across days of the week; across months of the year; and before & after a particular date (e.g., before & after my son was born). (8) shows the “days of the week” comparison, demonstrating that steps on Saturdays were the highest among other days of the week.

Implementation

We implemented Visualized Self as a web-based application with a server backend. The backend handles user management (signup, login) and also functions as a data backend (all data users acquire from external data providers are stored both locally and remotely on the server). The frontend uses a local database that synchronizes with the remote database server, thus ensuring data persistency. It also takes a modular approach to connecting to external data services such as Fitbit or RunKeeper. Connections to these data services work via APIs and most external data services rely on the OAuth process for authentication.

Contribution

The nature of our study is open and qualitative; the focus is not to test the system but to collect contextualized feedback by introducing a new technology concept, which supports a subset of activities while leaving aspects of the design open. Please read more from the paper blow.

The key contributions of this paper are:

An understanding of how self-trackers explore personal data with timeline visualizations and temporal comparisons based on situated feedback.
Types of personal insights self-trackers identify from visual data exploration.
Opportunities and design implications for visual data exploration platforms to promote self-reflection.

Page updated

Google Sites

Report abuse