Analytics Forward 2019 - A Story of an Unplanned Un-conference Presentation

Every year Analytics Forward is one of my favorite conferences. This is my 4th year in a row attending.

Since I have been giving more lightning talks and last fall presented at Jupyter Day in the Triangle as well as Microsoft AI Bootcamp. I felt inspired to pitch a talk for Analytics Forward. Since Analytics Foward is a Barcamp style Un-conference, rather than a formal CFP process the organizers have a keynote speaker designated ahead of time but all of the other sessions are chosen by a voting process the day of the conference.

So a month before the conference I started to think about what I could do for a talk. At first I wanted to explore the data in my hobby project The Walks of Nathan Lowell; however the appeal of walking an audience through writing a scraper that goes through his blog, through his Flickr and joins those with his tagged tweets from Twitter did not seem so exciting. Instead, I thought about what I actually know. Well I know analysis, and I know that what has made me a good analyst has been understanding a specific industry and a specific company.

Early one morning I was reading through something on my blog feed in Feedly and my pitch statement came to me:

You have mastered the convolusional neural network.

You were doing adversarial neural networks before they were cool.

Your bar charts are liked by Tufte on twitter.

Even in 3rd grade you knew that the 3D pie chart Mrs Hazelet put on the overhead was just… wrong.

But no matter how good your data skills are if you don’t understand the domain, your analysis may fail.

I quickly jotted my thoughts down in Evernote thinking I would create my talk around this core idea. Unfortunately a few days later I caught the Flu. Not just a 24-hour bug, but full on body aches, 103.8 degree fever, don’t get out of bed for 2 days Flu. After 2 weeks of constantly feeling exhausted, wondering if I could even attend a full day conference without falling asleep after 3 hours. I abandoned my thoughts of giving a talk.

Luckily I did start to feel significantly better the week of the conference. At least I would be able to attend.

Saturday morning was chaotic in our house. I was getting ready to leave for the conference. My wife was trying to get out the door to a volunteer event. The little guy just wanted to watch Saturday morning cartoons… I barely made it on time.

Just before the opening remarks I commented to my former co-worker that I had decided not to pitch a talk. That I had nothing prepared. She commented that one of the organizers has nudged her to pitch an open session as he was worried that they would have trouble filling all the spots.

Eric was about halfway through his remarks. My tablemate Xan Gregg grabbed a marker to fill out his talk sheet. The line for pitches started to form. We had an extra sheet of paper on our table… I figured why not. There were a lot of good speakers in line, I had my pitch from weeks before, my talk was mostly in my head if I made it I could create a few slides.

I was near the end of the line. There were 24 total pitches. I pulled out my phone while in line and frantically searched for my note. I found it when there were about 6 people in front of me in line. I usually don’t get nervous in cases like this. I can usually take 10 seconds and relax before I start, not this time. I gave my pitch, stumbled over a couple of words as my nervous tongue didn’t wrap around ‘convolusional.’ I noted that I had a 20 minute talk and the rest would be a discussion around Domain Expertise. Done.

Soon voting began. After about 10 minutes I glanced at my sheet. I had 14 votes. That is a lot. I quickly found 2 of the organizers and mentioned to please put me late in the day, that I needed to write my talk. Both of them know me pretty well. I doubt they were surprised. I sent a text to my former co-worker.

So much for not pitching. Working on my slides now.

Everything came together quickly. I decided to use Jupyter with RISE for my slides, knowing they would be easy to share on Github.

I was done in a little over an hour. I did break at the end of the first session to mingle with folks and have a cup of coffee. I pushed my Jupyter Notebook to Github and considered myself done.

My presentation slot was at 2:30, so I was able to attend a session in the morning and the keynote with the confidence that my talk was ready to go.

I ended up with 22 people attending. A good number in my opinion. The other 2 talks at the time were the amazing Dr. Zedy Ortiz, a refined speaker and well known in the Data Science community, and Melinda Thielbar one of the founders of the host organization Research Triangle Analysts.

I feel like my talk went well. I delivered what I said I would in the pitch. It is hard for me to be objective beyond that.

So will I pitch next year? Yes, I think I will. Maybe by then my ‘Walks of Nathan Lowell’ project will be refined enough to be interesting.

Analytics Forward Homepage
Edward Tufte on Twitter
Jupyter with RISE slideshow by Mike Driscoll

Written on March 9, 2019