Philly Twitter is negative, but not most negative, according to Data Jawn

The reboot of Data Jawn, the Philly-centric data science event that took place Wednesday at the Science History Insitute, came peppered with deep thoughts, good memes and hard data.

Out of over a dozen speakers, perhaps the most striking was Magento’s Ben Garvey, who analyzed 1.5 million geo-tagged tweets in answer the long-running question: Is the “negadelphia” myth true? Are Philadelphians inherently negative?

Garvey captured over 100,000 tweets from each of the 13 biggest cities in the country and ran them through a speech sentiment analysis tool. After all that data-parsing, a gut-wrenching conclusion: Philly Twitter ranks 10th least positive among big cities in the U.S.

“Philly is one of the least positive-tweeting cities in America,” Garvey said, signaling our city’s score of 0.72. Los Angeles tops the list as least negative city with a score of 0.99.

Even though the analysis focused on 2017, Garvey took additional sampling to take a look at how the Eagles Super Bowl win and subsequent victory parade affected our spirits.

“It looks pretty clear that the Super Bowl on Feburary 4th and the Parade on February 8th did in fact raise the tweeting sentiment for Philadelphians,” Garvey said. “However, this boost in positivity doesn’t last very long and it sank below 2017 levels by mid February.”

The data-driven confirmation of Philly Twitter’s negative reputation was just one of many knowledge bombs set off Wednesday at the event, organized by a handful of Philly companies like CompassRed and Stitch.

Here’s just part of what we thought of on the way home:

Face the actual problems in data science

Vicki Boykis, manager of data science and engineering at CapTech Ventures, said growing fears around human obsolescence in the age of artificial intelligence and machine learning are misguided.

“The hard problems still remain human problems,”said Boykis. “Actual problems include: moving and keeping data in the cloud, security and data privacy and model intent.”

Move past the hype of what some say are challenges of machine learning and AI. Instead, focus on three real ones, says @vboykis

☁️ Keeping data in the cloud
⚠️ Security and data privacy
🤖 Neural nets that think like humans#datajawn pic.twitter.com/Y9rczrU6Wr

— Technical.ly (@Technical_ly) June 13, 2018

The nonprofit world needs data science in their own terms

Sam Chenkin, director of consulting at TechImpact, reminded a roomful of data scientist that, when working with nonprofits, its best to meet them where they are in terms of capability and resources.

“These are small organizations doing big, important work,” Chenkin said. “Barriers include lack of investment, time and energy.”

To help organizations deploy data science solutions, it’s best to start by using what tools and methods are already in place, built self-service tools that limit the need for extra work and demonstrate the potential impact data could have to their mission.

The pharmaceutical industry is not very good at creating pharmaceuticals

A whopping 90 percent of the medicines the pharma industry tries to develop fail, with an average $2.6 billion spent in R&D for each approved drug. But data science is already in place to help those numbers go down, said Bonnie Kruft, director of data science strategy at GSK’s R&D Data Center of Excellence.

Scientists at the global company are using neural networks to identify patterns and features correlated closely with failure.

Data from Earth is awesome, useful and free

Andrew Pawloski, a Philly-based senior DevOps engineer at D.C.-based Element84, reminded the audience that data from Earth, as provided by NASA, is a national asset and therefore free for anyone to use.

“There’s not period of exclusive access and all users will be treated equally,” he said. Here’s one of many online tools to start exploring.

HR could benefit from looking at similar data sets as marketing and sales

For EmployeeCycle CEO Bruce Marable, the connection should be easier to make: human resources should look at data sets like marketing and sales does in order to track progress, pain points and opportunities for growth. (That line of thinking is a selling point for Marable’s company, which makes a dashboard for HR professionals.)

The ever-dapper @BruceMarable says #HR departments should track similar data sets as marketing and sales:

📣 Efficiency of advertising channels for their job postings

📆 Time it takes to fill positions

🗂 Size of funnel (candidate pool)#datajawn pic.twitter.com/WFES1WSeoZ

— Technical.ly (@Technical_ly) June 13, 2018