There Will Be Math

• Distance Formula
• \text{Given points}A\left(x_{\mathrm{a}},y_{\mathrm{a}}\right)\text{and}B\left(x_{\mathrm{b}},y_{\mathrm{b}}\right)
• \text{Distance}d\text{between}A\text{and}B\text{defined as}
• d=\sqrt{{\left(x_{\mathrm{b}}-x_{\mathrm{a}}\right)}^2+{\left(y_{\mathrm{b}}-y_{\mathrm{a}}\right)}^2}

Grand Meetup 2013

Grand Meetup 2012

Grand Meetup 2012

Grand Meetup 2013

Grand Meetup 2012

Grand Meetup 2013

Grand Meetup 2012

Grand Meetup 2013

Grand Meetup 2012

Grand Meetup 2013

A Pattern

Gather Data

Gather Data

• Normalize years, so same photo dimensions
• Map coordinates for each face, in each year
• Calculate distance moved from one year to next
• 93 people identified in both 2012 and 2013 photo

• Median Distance Moved: 278 pixels

Explore Data

Top Stationary
Paul Sieminski:	22 pixels
Philip Arthur Moore:	24 pixels
Paul Maiorana:	38 pixels

Top Movers
Kevin Conboy:	871 pixels
Nikolay:	863 pixels
Westi:	852 pixels

Explore Data

What should we expect?
Hard to tell, is this normal or not?

GM Photo Simulation

Explore Data

• Ran simulation of 10,000 Grand Meetup Photos, randomizing spots
• Average distance was 419 pixels moved
• Actual median distance moved 34% less than expected! So yes, likely a person tends to choose a similar spot.

Height Bias

Don't people line up by height, this is all bogus!?

Chairman of Height

Calculate Again just X-Axis

• Median X Distance Moved: 269 pixels
• Simulation for X-Axis: 384 pixels
• Difference: 30%

Conclusion

People might have a tendency to stand in the same spot.

More analysis is needed. A spot may be determined by direction people fill in. Some might be consistently late and the closer spots are filled.

20 more years of photo analysis will help get statistical accuracy.

What I learned

• Learned a bit of MathML for formula
• Played with Pandas and iPython Notebook
• Played with D3.js to create simulation
• Studied everyone's faces in prep for meetup

Slides built in HTML using dzslides
All photos used without permission