H517 Visualization Design, Analysis, & Evaluation Week 15: - PowerPoint PPT Presentation

H517 Visualization Design, Analysis, & Evaluation Week 15: Example Vis Research projects Khairi Reda | redak@iu.edu School of Informa5cs & Compu5ng, IUPUI

Scientific visualization techniques BactoGeNIE (genomics) BMC Bioinformatics, 2015

Human Factors in Visual Analytics Effects of Display Size on Insight ACM CHI’14, CHI’15

Visualization for Science Education Communicate science narra5ves with interac5ve visualiza5on and real-5me simula5on Rain Table, 2009-2011 American Museum of Natural History Field Museum, Royal Ontario Museum

Administrativia… • Next week : final presentation • Demo a working version of you visualization • Just bring up the vis; no need for slides • Presentation time: 6 minutes +1 min Q&A • Upload your vis to a public URL so it can be accessed from class computer • Final deliverables due in Canvas: Dec 11 (midnight) Extra o ffi ce hours Thursday 1pm - 3pm Friday 12pm - 1pm

‘Omics Data • Genome sequencing costs have decreased faster than Moore’s Law • 1000s of complete genomes sequences • New opportunities for comparative genomics research Do existing genome analysis tools scale?

Fong et al. PSAT: a web tool to compare genomic neighborhoods McKay et al. Using the Generic Synteny Browser (GBrowse_syn) . of multiple prokaryotic genomes. BMC bioinformatics 9:1 (2008) Current protocols in Bioinformatics Hoboken, NJ, USA: John Wiley & Sons

Ensemble encoding Y X

Effects of Display Size on Insight ACM CHI’14, CHI’15

=? + + visualization user data space Is temporal-separaJon of data detrimental to data analysis?

Are people be_er at analyzing informa5on, when informa5on is distributed spaJally?

“Space to think” Andrews et al., CHI’10

How many insights users can come up with? And what is the nature of these insights? abc… efg… 123… 456… small display large display (temporal-separa5on) (spa5al-separa5on)

Study Design • Volunteer graduate students were recruited to par5cipate. Had basic knowledge in data analysis and experience with big displays • Between subject design: par5cipants split evenly between two condi5ons ( small vs. large ) • Task: visually explore and analyze crime pa_erns in Chicago over the last decade (~ 2.8 million data points) • Think-aloud protocol • Open-ended exploraJon: for a maximum of 2.5 hours

Visualization Interface detail overview map magic lenses 2012 2006 2009 weapon narcotics violations

Two experimental conditions large small 3 x 4 panels 13 x 4 panels 12 Megapixels 54 Megapixels 4.5 X 40° FOV 190° FOV

Analysis • Insights • Observa5ons • Hypotheses • Insight breadth score: 1 … 5 1 — “ I can see a lot of non-serious crimes in downtown Chicago. ” 5 — “ A lot of people in the north-side are doing drugs, but they’re not fighKng - there are much fewer deaths resulKng from the narcoKcs trade [compared to the south-side] ” • ExploraJon Jme: how much 5me par5cipants choose to spend on the task

Exploration time 40 min extra time spent on exploration time task with the large display minutes p < .01 results

Reported insights No significant difference 74% more observations observation rates observations reported with the large in rate of insight acquisition display observation / minute insight / min of analysis small large 5 1 2 3 4 * * * * * * * distribution of breadth scores p < .05 𝝍 2 (4,1327) = 263.3, p < .001 results

Insights over time commutative insights cumulative insights small large results minutes into activity

Are large displays better for exploratory data analysis? Big displays help users… • discover more insights • integrate different pieces of informa5on • engage with and spend more 5me on the analysis On the other hand • Big displays could discourage a narrower, more focused reading of the data • Big displays will increase the 5me of analyses

Visualization for data- oriented storytelling

Informal Science Education

Flow of surface water • Water flows toward other water • Rivers curve due to the Earth’s spin • Water flows south no ma_er what

Rain Table Real-5me simula5on and visualiza5on of water flow

Set C Set A Set B Set D X Y X Y X Y X Y 10.0 7.46 8.0 6.58 10.0 8.04 10.0 9.14 8.0 6.77 8.0 5.76 8.0 6.95 8.0 8.14 13.0 12.74 8.0 7.71 13.0 7.58 13.0 8.74 9.0 7.11 8.0 8.84 9.0 8.81 9.0 8.77 11.0 7.81 8.0 8.47 11.0 8.33 11.0 9.26 14.0 8.84 8.0 7.04 14.0 9.96 14.0 8.10 6.0 6.08 8.0 5.25 6.0 7.24 6.0 6.13 4.0 5.39 19.0 12.50 4.0 4.26 4.0 3.10 12.0 8.15 8.0 5.56 12.0 10.84 12.0 9.13 7.0 6.42 8.0 7.91 7.0 4.82 7.0 7.26 5.0 5.73 8.0 6.89 5.0 5.68 5.0 4.74 mean 9.0 7.5 9.0 7.5 9.0 7.5 9.0 7.5 variance 11.0 4.12 11.0 4.12 11.0 4.12 11.0 4.12 regression y = 3 + .5X y = 3 + .5X y = 3 + .5X y = 3 + .5X R 2 =0.82 R 2 =0.82 R 2 =0.82 R 2 =0.82

Set A Set B Anscombe’s Quartet Set C Set D

Statistics / machine learning + Visualization

Please complete course evaluation! https://coursequestionnaire.iu.edu/blue Or tinyurl.com/viscourse9