Harnessing the Power of Visualization for Human...

Post on 25-Jul-2020

0 views 0 download

Transcript of Harnessing the Power of Visualization for Human...

Bongshin Lee

Microsoft Research

Harnessing the Power of Visualizationfor Human-Data Interaction

The Rise of Self-Tracking

• 259,000 mHealth apps listed on major app stores (2016)1

• Nearly 25% of Americans own a wearable (2016)2

1 Research 2 Guidance, mHealth App Developer Economics 2016.2 Rock Health, 50 things we now know about digital health consumers

Promises

Personal Data Self-Knowledge Self-Improvement

?

구슬이 서말이라도 꿰어야 보배

It takes more than pearls to make a necklace.

Nothing is complete unless you put it in a final shape.

Information Visualizationthe use of computer-supported, interactive, visual representations of abstract data to amplify cognition [Card et al., 1996]

I explore innovative ways to help people understand and communicate their data

leveraging visualization.

Why visualize data?

Anscombe's Quartet

• Mean of the x values = 9.0

• Mean of the y values = 7.5

• Equation of the least-squared regression line: y = 3 + 0.5x

• Sums of squared errors (about the mean) = 110.0

• Regression sums of squared errors (variance accounted for by x) = 27.5

• Residual sums of squared errors (about the regression line) = 13.75

• Correlation coefficient = 0.82

• Coefficient of determination = 0.67

Anscombe's Quartet

Anscombe's Quartet

• X Mean: 54.26

• Y Mean: 47.83

• X SD: 16.76

• Y SD: 26.93

• Correlation: -0.06

[Matejka & Fitzmaurice, CHI 2017]

Datasaurus Dozen

128176875613897654698450698560498289809858458224509856458945098450980990910302099059595957725646750506789884578980982167765487636490856091299856049828267629809858458224509856484582245098564589450984509809435859

How many 3’s?

128176875613897654698450698560498289809858458224509856458945098450980990910302099059595957725646750506789884578980982167765487636490856091299856049828267629809858458224509856484582245098564589450984509809435859

How many 3’s?

Is a red circle present?

Is a red circle present?

Is a red circle present?

My Publications

http://www.wordle.net

My Publications

My Publications

HCIHCI

Visualization

Visualization for Machine Teaching

Natural Interactionfor Visualization

Data-DrivenStorytelling

Personal Vis &Self-Monitoring

Research Goal

https://recoveringengineer.com/leadership-skills/new-supervisor-skills-people-must-feel-empowered

Analyze ShareCollect

Human-Data Interaction

Collect ShareAnalyze

TouchPivot [CHI 2017] Visualized Self [PervasiveHealth 2017]

How to help novice users visually explore data?

Powerful way to help people gain meaningful insights from their data

Visual Data Exploration

Visual Data Exploration

Data Insight

But, it is difficult for novices to perform visual data exploration.

Grammel et. al, How information visualization novices construct visualizations. [IEEE TVCG, 2010]

TouchPivotBlending WIMP & Post-WIMP Interfaces for Data Exploration on Tablet Devices [CHI 2017]

Bongshin Lee

Microsoft Research

Sehi L’Yi

SNU

Jaemin Jo

SNU

Jinwook Seo

SNU

Data Transformation & Visual Mapping

State Gender Population

Colorado Male 700

Colorado Female 900

Utah Male 400

Utah Female 300

Data Transformation

Pivot by Gender

GenderSUM

(Population)

Male 1,100

Female 1,200

0

500

1000

1500

Male Female

Population

Visual Mapping

Shelf Configuration Interfaces

PivotTable(Microsoft Excel)

Tableau

Where should I put this column?

To alleviate novices’ hurdles in visual data exploration

leveraging pen and touch interactions

Research Goal

TouchPivot Demo Video

https://www.youtube.com/watch?v=Q6quofDiO7I

Favor simplicity over flexibility

Provide tight coupling between data and visualization

Suggest appropriate visualizations

Combine pen & touch with WIMP interfaces

Design Rationale

How do people reflect on their own self-tracking data?

Eun Kyoung Choe

UMCP

Understanding Self-Reflection:How People Reflect on Personal Data through Visual Data Exploration [PervasiveHealth 2017]

Bongshin Lee

Microsoft Research

Haining Zhu

PSU

Dominikus Baur

Nathalie Henry Riche

Microsoft Research

Challenges with Personal Data ExplorationData is scattered across multiple platforms [Li et al., 2011; Choe et al., 2014.]

People don’t know what to do with the data [Choe et al., 2014; Epstein et al., 2015; Lazar et al., 2015.]

Analyzed 30 video recordings of QS presentations

[IEEE CG&A 2015]

Eun Kyoung Choe m.c. schraefelBongshin Lee

Types of Personal Insights

1. What I did

2. How I did it

3. What I learned

Detail

Self-Reflection

Trend

Comparison

Correlation

Data Summary

Distribution

Outlier

Visualization Insights

(74%)

(51%)

(36%)

(35%)

(11%)

(9%)

(6%)

(2%)

Data Integration from Multiple Sources

Data Summary

Trend Comparison

Interactive Data Exploration

Study Session

“I think that was soon after my surgery and that maybewould make sense cause I’d have to get up to takemedicine and maybe being restless or something.” [P8]

Visual data exploration Contextual information

Insight Gaining Pattern #1

External context

Visual data exploration Contextual information

Insight Gaining Pattern #2

External context

Question (Hypothesis)

Visual Data Exploration: Comparison by time segmentation

Question Did changing jobs affect my weight?

P1: [entering Sept 15, 2015 to compare his weight before and after this date]

Researcher: “Why Sept 15?”

P1: “That's kind of around the time I changed jobs. I was wondering if there wasanything interesting there.”

Support personal data exploration on mobile environment

Incorporate system-driven insights

Capture and share interesting questions and insights

Going Forward …

Collect Analyze Share

How to help people communicate their insights?

ChartAccent [PacificVis 2017]

Bongshin Lee

Microsoft Research

Matthew Brehmer

Microsoft Research

Donghao Ren

UCSB

Eun Kyoung Choe

UMCP

ChartAccentAnnotation for Data-Driven Storytelling [PacificVis 2017]

Tobias Höllerer

UCSB

“the annotation layer is the most important thing we do . . . otherwise it’s a case of here it is, you go figure it out.”

—Amanda Cox, The New York Times Graphics Editor

Current annotation support is limited

ChartAccent Demo Video

https://www.youtube.com/watch?v=8ogkSBlE2f0

Annotation Target Annotation Form

Annotation Design Space

Annotation Form

Text Shape Highlight Image Combined

Annotation Target Type

Annotation Target Type

chartaccent@gmail.com | @chartaccent

Examples, Tutorial, and Survey

Open-source project -- https://github.com/chartaccent

ChartAccent.github.io

ShareAnalyzeCollect

How to support people’s diverse tracking needs?

OmniTrack [UbiComp 2017]

A Flexible Self-Tracking Approach Leveraging Semi-Automated Tracking[UbiComp 2017]

Bongshin Lee

Microsoft Research

Jae Ho Jeon

Kakao Corp.

Young-Ho Kim

SNU

Jinwook Seo

SNU

Eun Kyoung Choe

UMCP

The Rise of Self-Tracking

• 259,000 mHealth apps listed on major app stores (2016)1

• Nearly 25% of Americans own a wearable (2016)2

1 Research 2 Guidance, mHealth App Developer Economics 2016.2 Rock Health, 50 things we now know about digital health consumers

Most self-tracking apps provide little or no flexibility.

Imagine how you might record your reading activities

Diverse Tracking Needs

texttextstar

Book Reviews

TitleAuthorRating

Book Reviews

texttextnumbertext

TitleAuthorPagesReview

Reading Logs

TitlePage FromPage ToDate

textnumbernumberdate

Challenging to find an existing app that perfectly suits one’s tracking needs

Research Goal

To support people’s diverse tracking needs through

a flexible self-tracking system

Semi-Automated Tracking

Reduced mental load

Better accuracy (depending on the data)

Cumbersome to wear (wearable sensing)

Reduce engagement with data

+

+

--

Engagement with data

Increased self-awareness

Flexibility of choosing target behaviors

Some data can only be tracked manually

High capture burden

Compromised data accuracy

+

+

+

+

--

Fully manualtracking

Fully automatedtracking

Choe et al., Semi-Automated Tracking: A Balanced Approach for Self-Monitoring Applications [IEEE Pervasive Computing 2017]

Semi-Automated Tracking

Fully manualtracking

Fully automatedtracking

Paper DiaryEmbedded Sensing

Semi-automated Tracking Spectrum for Sleep

Balancing Burden with OmniTrack

Automated Capture of Sleep Duration

Manual Capture ofSubjective Sleep Quality

Fully manualtracking

Fully automatedtracking

Semi-automated Tracking Spectrum for Sleep

OmniTrack Demo Video

https://www.youtube.com/watch?v=zRIuENNRjEM

Experiment Design

Tracker Design 1

Tracker Design 2

Group A

Group B

Data & Experiment Management

Progress Monitoring Intervening

Create data collection tools without programming

OmniTrack ResearchKit

Researcher Participants

Data

Trackers

Interested in conducting a diary study/ESM using OmniTrack ResearchKit?

Contact

Personalized Self-Reports

Self-Report + Fitness Device Tracking

Multi-Device Data Collection

omnitrackapp@gmail.com

.github.io

omnitrackapp@gmail.com | @omnitrack_app

Detailed Information, Supplementary Materials

OmniTrack will be open-sourced soon!

Analyze ShareCollect

Human-Data Interaction

Research Avenues

Mobile Data Visualization

Patient-Clinician Communication

Visualization Literacy

Analyze ShareCollect

One Last Thing

Work with visualization researchers

Thank You!

Analyze ShareCollect

bongshin@microsoft.com | aka.ms/bongshin | @bongshin