Global Nerdy

Here’s a new regular feature on Global Nerdy: 3 things, in which I’ll take a programming or technology topic and post three things that I found of interest on that topic. Today’s topic is a hot one: data science, which often goes hand in hand with Python.

The best data science course on the internet, according to someone who’s created an online master’s in data science

In the process of creating his own data science master’s program based on online resources, Udacity content developer David Venturi created a review-driven guide that recommends the best courses for each subject within data science, including:

• If you want to learn Data Science, start with one of these programming classes
• If you want to learn Data Science, take a few of these statistics classes

And now, he’s got an article with a near-irresistible title (well, it’s near-irresistible if you’re the sort of person who reads this blog): I ranked every Intro to Data Science course on the internet, based on thousands of data points.

I’ll spare you the suspense: the top-ranked course in his list is Kirill Eremenko’s Udemy course, Data Science A-Z™: Real-Life Data Science Exercises Included.

I like to think of Eremenko as the Tim Ferriss of data science; both bill themselves as lifestyle entrepreneurs and have an empirical approach. If his name sounds familiar, you may have heard of his site, SuperDataScience, or his podcast, SuperDataScience Careers Podcast, or his set of Udemy courses related to data and statistics.

Gauge Donald Trump’s mood by using Python to do sentiment analysis on his tweets

If you’ve been meaning to get started with Python and text analysis, here’s a fun way to do it: performing textual analysis on the tweets of the world’s first internet-troll-turned-head-of-state, as shown in this article, Sentiment analysis on Trump’s tweets using Python.

You’ll use:

• Tweepy to access the Twitter API and fetch Donald Trump’s tweets
• pandas to perform data analysis on those tweets
• NumPy, matplotlib, and seaborn to perform statistical analysis and create visualizations
• TextBlob to do sentiment analysis

To do this exercise, it’s strongly recommended that you use the Anaconda distribution of Python, as it comes will all sorts of useful tools specifically for data science.

Python podcasts

And finally, if you want to immerse yourself in Python while driving, walking, at the gym, or while away from the computer, you’ll want to check out Dan Bader’s Ultimate List of Python Podcasts.

Here’s a four-minute video where Trey Steinhoff, one of the people behind the recent Hack Hospitality hackathon in St. Pete, talks to me about Sourcetoad’s involvement with the event.

Hack Hospitality brought together two groups that normally don’t overlap — techies and the hospitality industry, both from the Tampa Bay area — with the purpose of coming up with applications to solve problems that the hospitality industry faces.

In the video, we talk about:

• Why Sourcetoad was involved with the hackathon. The short answer: Because hackathons are a key part of a healthy tech ecosystem, and a healthy tech ecosystem is good for the company. Tampa Bay has the right elements to become the Austin of Florida, and we’re doing what we can to see that come about.
• What Sourcetoad does. Sourcetoad is a software development shop that doesn’t just build software for clients. We take on “the hard stuff” — the highly networked, analytical, data-crunching, enterprisey, internet of things-y development work that confounds other software development organizations.
• What Sourcetoad brought to the hackathon. Since one of our clients is a cruise line, we brought cruise ship-related datasets and challenges. The datasets contained all sorts of information specific to cruise ships, while our challenges were about real issues facing the cruise industry.
• What was special about Hack Hospitality: It was a confluence of greatness — great organizers, a great space (The Station House and Iron Yard are by far the most beautiful and comfortable hackathon venues I’ve been in), the best catering at a Hackathon that I’ve experienced, thanks to Ciccio Restaurant Group (the people behind Green Lemon, Daily Eats, and other places) and Saltblock Hospitality Group, and of course, the enthusiastic Tampa Bay tech community.

Betty Pierce speaks at the start of Women Who Code Tampa’s “All About Agile” meetup.
Click the photo to see it on full size.

Tuesday’s Women Who Code/Tampa Bay Agile meetup at AgileThought’s offices featured Julee Bellomo and Wendy Vigre introducing the concepts of agile software development to the audience of Tampa Bay’s most signed-up-for event of the evening.

Betty Pierce speaks at the start of Women Who Code Tampa’s “All About Agile” meetup.
Click the photo to see it on full size.

In the spirit of the Agile Manifesto’s “individuals and interactions over processes and tools” and “working software over comprehensive documentation” philosophies, Julee and Wendy wisely chose to demonstrate agile concepts by having attendees partake in activities rather than sit through a lecture.

Julee Bellomo presents at Women Who Code Tampa’s ‘All About Agile’ meetup.
Photo by Anitra Pavka.

We went through a number of exercises, one of which had everyone in the group stand in a place in the room that signified where they stood on the four spectrums shown below:

Julee Bellomo presents at Women Who Code Tampa’s ‘All About Agile’ meetup.
Photo by Anitra Pavka.

Those of you who are already familiar with the Agile Manifesto know that its creators favor the items on the left side over those on the right. It should be noted that they do believe that the items on the right side have value; it’s just that given each pair and the requirement to choose just one, they’d prefer the left one.

I was getting pizza during the point when we were asked to situate ourselves on the working software / comprehensive documentation spectrum, and the Hawaiian pizza  was near the comprehensive documentation extreme. When asked why I favored comprehensive documentation over working software, I replied “Because documentation doesn’t have to actually compile, or even work. You can often get away with simply submitting it.”

(I will always gladly play devil’s advocate for pizza with pineapples on it.)

The big exercise of the evening is one that agile coaches love: the agile coin exercise. They love it because it demonstrates something that seems counterintuitive to a lot of people: you can get things done more quickly, efficiently, and well by:

• Doing work in smaller units and delivering that work more often (instead of doing it in larger units, delivered less often), and
• Letting the team work together to come up with their own ways to improve how the work is done.

We divided ourselves into groups of 9, and in each group, there were three roles:

• 4 workers, whose job was to do the “work”.
• 4 managers, one for each worker. Their job was to use a stopwatch to time how long it took their worker to do the “work”.
• 1 HiPPO (HiPPO is short for “Highest Paid Person in the Organization”). Their job is to use a stopwatch to time how long it took for all the workers to do all the “work”.

In the game, the “work” was to flip 20 pennies to the opposite side. Each worker would do the “work”, then pass the pennies to the next worker, who would then do the same “work”, then pass the pennies to the next worker, and so on. The “work” would be completely done once all the workers in the group had done “work” on the pennies.

We had a number of iterations of the game:

1. In the first iteration, each worker had to flip all 20 pennies before passing them to the next worker. This took most groups around 70 to 90 seconds.
2. In the second iteration, each worker could flip 5 pennies, pass those flipped pennies to the next worker, flip the next 5, pass them to the next worker, and so on. This took most groups around 40 to 60 seconds, even though individual times grew slightly.
3. In the final iteration, the groups were free to decide how the work would be distributed. Many adopted the approach of flipping 2 pennies at a time, then passing them to the next worker. Individual times grew slightly more, but most groups’ aggregate times dropped to about 30 to 40 seconds.

By the end of the exercise, you can see how breaking tasks into smaller units and allowing the team to devise their own optimizations boosts throughput. That’s the general idea behind agile processes, no matter what they are.

If you’d like to try the agile coin exercise, there are a number of descriptions online, including this one, which was followed up with an article about some variations on the exercise.

You may also be interested in Remotely Flipped, an online version of the exercise that you can play with remote friends (it runs on Heroku, so give it a little time to start up). For those of you who’d like to know more about its implementation, check out this article on the blog of Agility Feat, the people who made it; it was developed using PubNub, Clojure, and ClojureScript.

You use lots of these in many agile methodologies.

The meetup closed with a quick overview of Scrum and Kanban, as well as Julee’s assertion that there’s no such thing as “Scrumban” (a term people use for hybrids of the two). She says that’s just “Kanban done right”.

I’m going to leave the differences between Scrum and Kanban as an exercise for the reader, but I’ll post this graphic as a way to help you get started: