A Quest for High Fidelity Activity Tracking

From Personal Science Wiki
Jump to navigation Jump to search
Project Infobox Question-icon.png
Self researcher(s) Jamie Williams
Related tools iPhone, Moves
Related topics Productivity, Sports and fitness, Heart rate, Sleep, Activity tracking, Social interactions, Location tracking

Builds on project(s)
Has inspired Projects (0)
Show and Tell Talk Infobox
Featured image A-quest-for-high-fidelity-activity-tracking.jpg
Date 2015/06/18
Event name 2015 QS Global Conference
UI icon information.png This content was automatically imported. See here how to improve it if any information is missing or out outdated.

A Quest for High Fidelity Activity Tracking is a Show & Tell talk by Jamie Williams that has been imported from the Quantified Self Show & Tell library.The talk was given on 2015/06/18 and is about Productivity, Sports and fitness, Heart rate, Sleep, Activity tracking, Social interactions, and Location tracking.

Description[edit | edit source]

A description of this project as introduced by Quantified Self follows:

Jamie got interested in tracking his time 15 years ago when he was a post-doc. He felt unproductive and decided to record the amount of time he was actually getting work done. Seeing the actual data gave him a sense of control but it became tedious so he stopped. In this talk, he shares the project that he has been working on to visualize a timeline of his daily activities and explore his habits through data visualization. He also provides a background of how the project has evolved and what tools he has built to so far.

Video and transcript[edit | edit source]

A transcript of this talk is below:

Jamie Williams A Quest for High Fidelity Activity Tracking

Hi, I’m Jamie Williams. Today I’d like to share a project I’m working on to visualize a timeline of my daily activities. I’ll give some background on how the project has evolved and show you what I’ve built so far. Increasingly, many aspects of our lives are leaving a digital trail of data. These different data sources are typically isolated from each other. One of the current big challenges of the QS community is to build tools to pull all this data together. And we’re seeing more and more apps and services sprouting up to fill this need. We just heard from Evan on Data Sense, and Flux Stream, Gyroscope, Serveit, and Zenobase are all represented here at the conference, each with their own unique vision. I’m fascinated on this problem and working on a personal project to aggregate my own data together to paint a picture of how I use my time. My current focus is on visualizing all my tracking data on a unified timeline. I first got interested in tracking my time back when I was a post-doc 15 years ago. I felt unproductive and decided to record the amount of time I was actually getting work done. I did this using a simple form pilot app and found it to be very helpful. Seeing the actual data gave me a sense of control. I became less stressed and could see my time management improve over time. But once I got on track it was just too tedious to keep logging my time, so I stopped. Much later when the iPhone came out, I decided to climb this information curve a bit further by building my own time-tracking app. So I built an app that would let me toggle from one activity to the next with a single tap. With the Palm Pilot, I had focused on work time. But here I tried to log everything I was doing in detail. Needless to say, logging data like this turned out to be very tedious, and I gave up after about a week. But there’s still something very alluring for me about having a narrative transcript of your life. If it could be automated and effortless why wouldn’t you want the data? It's becoming increasingly effortless to track all aspects of our life, so why not try to aggregate and automate all this data. At the very least it's very interesting from a biographical or archival perspective, but my gut feeling that integrating these datasets together will unlock new insights into practical problems. This is a complex problem space with many directions to go in. From a time tracking perspective, I think an interesting long-term challenge is to devise an algorithm that can synthesize these different data feeds together to generate a single, narrative story. But for me in the near term, for me, a natural starting point is to build a visualization tool to explore the entire dataset in a unified view. Without such a tool if I wanted to explore the correlation of my heart rate to some other activity, like watching the penultimate episode of Game of Thrones, for example, it's not very easy because each source is disconnected. So my first design goal for this tool was to pull all the data together onto a uniformed timeline. Location data from the Moves app steps and sleep from Fitbit, iPhone usage from the moment app and when I'm watching television. Combined together, these data feeds start to reveal a narrative story of what I’m doing. So I’m trying to build a tool that lets’ me answer questions in different scales, and zoom down to a specific event; when did I go to sleep on a particular day, or zoom out to a broad overview level to see trends and patterns, or investigate anomalies contrary to the trend. We’re all used to mapping systems like Google maps, which let you move fluently between many orders of magnitude of scales, I wanted a tool that treats my timeline activities in the same way that the information it displays should adapt the timescale you zoom to. So this is where my project is currently. I’m trying to balance out a zoomable timeline that displays every single event or measurement, with charts showing trends of aggregated data, and also contextual information like a map. So what’s next? I’ve actually obtained all of my data from each of these sources, but so far have only incorporated a handful. I plan on tackling Fitbit heart rate data next as well as several others. A while ago I spent a few months building a proof of concept for indoor proximity tracking using Bluetooth low-energy beacons and concluded it was definitely doable, so I plan on revisiting that, so I can add it to my expanding symphony of data.


About the presenter[edit | edit source]

Jamie Williams gave this talk.