Guys, We Need to Talk

So, this happened. Which wasn’t all that surprising after this happened.

If you’re wondering why this blog got so quiet, maybe I’m just not up to being a woman on the internet these days. As a wise man once said, “If you’re tired, learn to rest not quit.”

It’s not up to me anyway. I’ve gotten a few questions about what I think “we” should do about a data science group in the Triangle “with a focus on men empowerment.” My official opinion is: I think data science in general is pretty male-empowered. I would point out that the other meetups in the triangle organized by women and for women don’t specifically exclude men, but if someone’s got to keep people out to feel safe, let ’em.

As for my opinion on what you should do: There’s been some pretty extensive research on what the “male empowerment” sect says and does and how they get there, and there are some basic consistencies:

  1. They feel dissatisfied with their lives–and don’t know what to do about it.
  2. They lack good mentors and a community. A men’s-rights group gives them a place to belong, and it is not all that different from any other radicalization of a dissatisfied young person.
  3. In many cases, it starts with a very real grievance, and a lack of support. I’m not saying there’s no support available. I’m saying that specific person does not have a support system and has no idea about how to get one. Someone might even say that people who are drawn to staring at a computer screen all day are less able to find emotional support than the average person, and so it’s not surprising they act out in a certain way.

So, if you want to know what I think you should do, this is what I think you should do:

If you’re a woman: come to RTA. You don’t even have to talk about math. We’re pretty fun generally. At least we think so.

If you’re a man (And you can stomach it): Join this group. Mentor these guys. Keep them from doing something we’ll all regret.

Then come to RTA and let me buy you a drink. You’re all welcome here, regardless of the skin you live in, the family you choose, or what you do with other consenting adults in your free time.

If you can’t stomach joining this group, you should still come to RTA. We’ve built a community of people that is fun and welcoming–at least we think so. If you’ve been to an RTA meeting, and you didn’t feel welcome, we’d like to have a conversation with you about your experience and try to fix it. We started this group to build the data science community in the Triangle, not break it into sections. That will be our mission as long as I’m president.

I hope we see you there soon!

Posted in Uncategorized | 2 Comments

Pretty Quiet Around Here

This blog is mostly a re-direct to other blogs where I post. If you want current information about Melinda Thielbar, you can:

Visit my LinkedIn page.

See What I’ve been Posting on the JMP blog.

Go to the Meetup page for Research Triangle Analysts , a 501(c)(3) educational organization of which I am president and co-founder.

Thanks for stopping by!

Posted in Uncategorized | Leave a comment

Applying for a DataKind Chapter In the Triangle

Sometimes, the best thing you can do is step back.

My call for partners to apply for a DataKind chapter in the Research Triangle was answered by some of the most talented people I’ve met. Four volunteers (plus me) stepped up. Each of them had an amazing resume, as well as a busy schedule. Ultimately, we picked a team of 3 that had the best combination of available time, technical skills, and experience with community outreach. That team of 3 held an online “application party” last night, where they pooled their immense resources to answer the application questions.

I did not make the cut–and that was great! I was once asked for advice on how to be a great data scientist, and after a few minutes thought, I said “Always be the dumbest guy in the room.” I do not regret following that advice–even when it means I don’t get to be out front.

I have no doubt the Research Triangle will be a great place for a DataKind chapter. If and when that happens, the leadership team we put together will do a great job with it, and I’ll be pleased to be a member.

Posted in Uncategorized | Leave a comment

Applying for a DataKind Chapter

DataKind is a really impressive organization that connects people with data skills with not-for-profit organizations who need help solving problems. They’ve already completed some great projects, and there’s a lot of momentum to do even more.

It’s a great idea, and they’re taking applications for new chapters in other parts of the company. Have data skills? Want to do some good in the world? Check it out.

Posted in Uncategorized | Leave a comment

The Trouble with Pie Charts

Data visualization experts will often show pie charts as an example of “bad” visualization. Xan Gregg (one of my colleagues at JMP) has posted the challenge “One Less Pie” for pie day, to show some better ways to visualize data.

But, you may ask, what’s wrong with pie charts?

Much smarter people have already answered this question, but let’s take a look at a “real” example from the recent “Transmedia Survey” (run by the unstoppable Team Thielbar/Dansky).

One of the first questions we asked in that survey was “Of the books you purchased last year, what percent were e-books?” We could show the results using a pie chart, like so:

Image

However, the pie chart loses the ordering. It’s not obvious in the resulting chart that “Less than 20 Books” is ordered next to “20 – 50”. A better visualization, like a share chart (or a stacked bar), shows the proportions without losing the ordering:

Image

This gets more important as you add information to the visualization. Maybe we want to see if people who play video games buy more, less, or the same number of e-books as people who don’t. Which visualization would you prefer for that question?

Pie chart?

Image

Or Share Chart/Stacked Bar?

Image

I prefer the stacked bar graph because it lets you compare the proportions across groups easily and quickly. The information is more compact, and it communicates a lot more than scanning across four different pie charts.

There are places where a pie chart can work. For example, the “Do you want cool stuff?” question has a small number of levels, and there’s no real order to them. A pie chart communicates who said “yes” and who said “no” pretty easily:

Image

As you might have guessed, the graphs shown are JMP graphs. The pie charts come from JMP’s Graph Builder, and the share carts are the default visualization in the Categorical Platform.

You can follow the further adventures of the pie chart on the JMP blog or by searching the hash tag #OneLessPie on Twitter.

Posted in Uncategorized | Tagged , | Leave a comment

E-Books, Paper Books, Transmedia: How Much is Content Worth?

I rarely see this kind of calculus in the publishing industry–particularly when it comes to fiction.* Let’s see if we can change that.

The E-Book, Paper Books, and Transmedia Content Survey is now open! This survey was designed using the Choice Experiment design of experiments platform in JMP, which is quickly becoming my second-favorite JMP platform (second to the one I work on, of course).

So, let’s collect some data and find out if we know as much as we think we know about books, publishing, and media in general. Please forward and share as you much as you like!

*You may or may not have heard that I used to write books, and the occasional short story. May data habit became too great, and I had to give it up.

OK, that’s a lie: I get terribly uncomfortably when the people making business decisions about my work aren’t asking questions like “How much can we sell this for?” or “What does it cost us to make this?” I bailed on writing like a rat deserting a sinking ship.

Posted in Uncategorized | Tagged , , | Leave a comment

Online Video Viewing Survey

Thank you to everyone who answered the Online Video Viewing survey. You asked if I would post the results. I decided it would be in keeping with the spirit of the experiment to record a video, which you can watch on my YouTube channel.

Standard disclaimer: This was a survey I ran for fun, using responses from my friends. This has nothing to do with my employer, and neither the survey design nor the sampling protocol was scientific. Any generalization made from these results would be highly suspect.

Sampling Design: Convenience sample. We put a link up on our Twitter and Facebook feeds and asked for responses. Richard’s Twitter and Facebook feeds tend to be populated by people from the entertainment industry. My Twitter and Facebook tend to be friends of mine and/or people interested in analytics.

Survey Design: Preliminary. We used a lot of free text responses in an effort to collect general information about how people watch videos. The best information was collected through the free text responses–answers to questions we never thought to ask.

Most surprising result: People seemed far more likely to click on videos they found in social media, while videos that were e-mailed to them were almost as likely to be “saved for later” as they were to be viewed or clicked immediately.  It’s an open question whether this is because of how people use social media verses email or a result of how the question was asked.

Least surprising result: Most people skip advertising content as quickly as possible when viewing online content.

Most interesting results: The lists of favorite movies and television series from childhood and current were a lot of fun. Complete lists (sorted by responses) are located here: BestTV and here: BestMovies. Feel free to use them as a “must view” list. I know I will!

Posted in Uncategorized | 1 Comment