2020 Year in Review – TableauFit

I’ve ran this blog for 5 years and never done a real “in review” post. Most years, one year rolls into the next and time is itself marked, but ultimately not significant. Why make a resolution in January when you can start in October and have less people at the gym when you’re the most out of shape?

I was glad to see the close of 2019. In addition to losing Kelly, my husband’s best friend died suddenly not even 3 months afterwards. The Queen had her Annus horribilis, and I had mine with 2019. Surely, 2020 had to be better…

I’d hoped to take time to refresh, to travel, and to push further into NLP and doing in-depth digs with data visualization. Here’s what actually happened:

I got a lot better with Alteryx.

I was hoping to expand my skills with NLP and APIs. Be careful what you wish for, because I certainly did this year, but not for the reasons I expected! Between taking on some private NLP projects and needing to get access to my child’s schoolwork via API, Alteryx proved to be a solid solution to getting to the data. I was able to parse open-text survey responses and provide real insight. I’m proud of this work, and some other things I’ve done this year.

Getting more comfortable APIs was certainly a goal, but I didn’t expect it to be to understand my child’s schoolwork. Learning Management Solutions are not designed with the parents in mind, and certainly not for having parents guide their kids through assignments. After trying and failing to use the provided user interface, I caved and started hunting through API documentation.

I want to be clear: I am not a coder by convention. Pure persistence is how I sorted it out. But now, I have a repeatable Alteryx flow and a dashboard, with more planned. Blogs incoming.

I spent too much time shaping data.

Part 1: All the data

When I started doing analysis well over a decade ago, ALL of it was hard. Getting the data was hard. Making visuals required knowing what you wanted (bar chart, obligatory pie chart in the right corner, and long table to parse) and spending oodles of time making sub-queries. Then you spent days printing things to PDF and sending it with written paragraphs of what was in the chart or – worse – had a whole meeting about it. People would circle numbers and marching orders were set for the week. Then, you repeated it all again. Data heroics were fun, until they became tiring.

Nowadays, the visualizing and sharing the insight is so much easier with Tableau. What gets sticky these days is the amount of disparate data we want to analyze together (there will be posts on this). Today, we want the universal answer (42), but we also need to know how that answer came about. I spent a lot of time this year bringing disparate data together both personally and professionally. Clients need data from a number of sources, and frankly, to do an honest analysis, I do too.

Part 2: Data lineage and integrity

You can’t bring in all the data in the universe and not provide clarity on its origins. We had this battle with Covid-19 visualizations, and we’ve had it in offices for as long as I can remember. But now, in 2020, we have jargon and dedicated branches for it – data lineage and data governance.

I spent some time this year integrating practices to highlight where data came from and allowing users to easily understand the impact. Beyond just getting all the data in the analysis, my focus was highlighting how it affected the analysis within the visualization. Call this visualizing uncertainty on steroids. Also another area where I need to build a demo and post.

Part 3: Prep Maturity

I can tell you exactly how much Prep has matured since it’s release. It’s gotten easier to do a number of tasks. I love that it shows you certain things and it makes profiling data a breeze. Do I still want more from Prep? Absolutely! Until it drives itself, I will always find a way to complain.

  • For best performance on CSV and such, send to Hyper first (this works in Alteryx too). I wish I didn’t have to do this step.
  • Never underestimate the power of a pivot, especially if you can pair it with a data dictionary. If you don’t have a data dictionary, use the flow to at least start its creation.
  • Building repeatable processes makes things so much easier.
  • Explicit (named) vs inferred (assumed) are key concepts to understand in designing processes. If you can infer based on field type or other attribute, then your ETL will scale much easier. The most explicit you must be, the less flexible your process generally ends up being. This point, too, probably needs to be its own post.

Part 4: Data Model

One of my favorite newer features in Tableau Desktop is the object or data model. I’ve learned to take all the disparate data from above and leverage the benefits of how it works to get all the data in and make it feel relational without being a Cartesian monster. Hooray accuracy! YAY client experience!

It also allows me to build more flexible sources if I’m not sure what shape I need for my analysis. Later, I can prune and transition. This feature has been key for a lot of my projects. No surprises here: at some point, I’ll build a reasonable demo and share.

Part 5: NLP

Do you want to do good NLP? Figuring out some level of Python or R seems to be a given. Don’t ask me how long it took me to figure out how to embed Python in Alteryx or how I’m still sorting out adding it to Prep (I will get there….). All of this gave me parts-of-speech, lemmas, sentiment, and so much more to do various analyses.

The more metadata we have around language, the better our analysis. We can create semantically resonant topics by focusing on nouns, identifying group identification with pronoun use, or understanding how the noun gets modified with verbs, adjectives and the like.

I did actually viz.

Beyond ETL, I made visualizations. It’s hard to believe as they’re so interspersed between ETL. I’m most proud of the Juneteenth viz I did with Candra McRae. In all the years I’ve used Tableau, I had never done a collaboration viz. For that viz, I did use an API (see above points), but I also talked to a real human (over the phone) and we made plans, removed things that were out of scope, found data, and figured out design…together. We also wrote a post on it together. I hope to do more collaborations.

I played with Stroop effects and made this demo before the stay-at-home orders. I made a few things that aren’t shared yet (like my school tracking dashboard) and I also helped my kid complete a few of his dashboards. He likes to build apps and has decided Tableau is a great prototyping tool. I also analyzed the analyses on Public around Covid-19. I’ll probably have one final post here.

Most of my work, however, occurred in the background. As usual, I did cool things with clients. I also participated in some Foundation work, which was great and challenging. More to come here as well.

I struggled with the world (again).

Kelly used to joke that your college major made up some type of deficiency. She didn’t understand the group, so sociology it was. I, too, was perplexed by the world and ASL interpreting helped me understand a bit of it. I’ll be frank: there’s still a lot that’s a mystery and probably a part of why ethics appeal to me – they present ideals about what we should do, but not often what we choose to do. Being brave has significant ramifications, as we’ve seen with Timnet Gebru and so many others.

Covid-19 brought about both a pandemic and an infodemic. Ethically, I struggled. Personally, I was coming off providing hospice and knew first-hand the pains of loss. We weren’t all affected the same. During the H1N1 outbreak, I worked for a provider-led organization focusing on providing language interpretation. We attempted to model what it could look like, so we could target our education campaign in a variety of languages.

While I made ethical calls, I won’t say I did enough and doubt my impact. I’ve tried to amplify others doing this work as well.

I signed a book deal.

While I read voraciously as a child, we had few books in the house. At one point, I even read the dictionary. Signing a book deal seemed more like a pipe dream than a potential reality.

But, here’s the thing: you find people to collaborate with that lift you up, that see the best in you and remind you of what you can do. Kelly did that for me, but so does Vidya Setlur. I’m proud to share author credits with her and will be heads-down writing a book in 2021.

I presented a lot.

When everyone else went remote, I finally had a reality that accepted the things I do best: working from afar. I hate the process of travelling. I’m a terrible flier, an extreme introvert, and I get every cold in existence. Artificial lighting exhausts me. Noise distracts me. I’m perfect for the remote experience. (No, really, book me!)

I took to a number of the TUGs, presenting on color, ethics, mobile design, and whatever other topic was relevant. You lose the feedback loop remotely, which is hard. But, I hoped I provided interesting ideas and maybe even a brief distraction.

I learned to make pizza.

Spend enough time with me and you’ll learn I hate to cook, but love to eat. I’m a chefless gourmand forced to learn to cook in a pandemic. Pizza is not my favorite food – far from it. Most pizzas come buried in cheese, offer far too few veggie toppings and end up greasy or hard to chew.

Folks, lemme tell you: the perfect pizza is a possible. You either make the dough, letting it mature for 24-hours, or you cheat and buy the raw starter. You cut all the ingredients – fine, long slices of red and green pepper, impossibly thin red onion curls, finely sliced portabella mushrooms, and whatever veggie protein your heart desires. Buy the pizza sauce, but get one of those fancy local brands with a minimally designed wrapper. Make the child’s pizza first. With what’s leftover, roll to around a 1/2 inch or less height. Place on parchment paper. Bathe with olive oil and garlic – use one of those fancy food paintbrushes. This will be flipped and be at the bottom.

On a recipe blog, I’d segue into some story about how this recipe came to be. It’d be 10 paragraphs long and have 7 advertisements. Each paragraph would be 2 sentences, but with all the ads and pop-ups, it’d feel longer. Trust me. This is a data blog, so I’ve made this recipe every week for 6 months and this is the sweet spot.

Prebake with nothing but the oil and garlic on it for about 5 minutes at 425 Fahrenheit (convection is ideal). The child’s fluffs up like a pita (no oil or garlic – bleh!), but ours just ends up with some bubbles. Remove, flip without burning yourself (this takes skill), and THEN top.

Resist the urge to bathe this in sauce. I love a saucy pizza, but too much creates goo and mess. For minimal cheesers, I recommend shredding your own cheese. Cheddar is surprisingly good here, as is gouda. The sauce is not only still visible at this stage, but only lightly covered with a nominal amount of cheese. I like to put mushrooms and protein first, then peppers, with onions last. Make this artful. Play with patterns.

I may have forgotten the cheese in this one, which is why it’s on top.

Bake for another 9 minutes. Here’s the other trick. Take it out. Cut. Spread pieces just a bit and then put it back in to crisp up the inside edges for another 2 minutes. No, this is not efficient in any way. Could you make a better pizza faster? I’m sure you could.

Why does pizza matter?

For me, it highlights the impact of the pandemic. I’m offline a lot more. I need that separation from the computer to be in this life, to spend time with my family (the reasons I do what I do), and to still find a way to savor things I enjoy. While I’m not a huge fan of pizza, I’ve found a way to make it tolerable for months on end. Reminder: I do this weekly.

Optimism is not being cheerful all the time. Michael J. Fox talks about optimism as gratitude. So much sucks for him, and he’s acknowledged it, but is finding ways to be grateful is key to happiness. I spent much of 2019 being grateful for some key last moments.

The myth introverts get sold is the deserted island is paradise. Without a community, without those forced face-to-face interactions, I have found myself adrift. While the above looks like a lot, it’s a small share of what I’d have done without a pandemic and without access to broad community support. I’m working and doing math with the kid. He zoom-bombs me often.

My hope in 2021 is not only the obvious productivity goals, but to find ways to be a better human, a more compassionate parent, and a more grateful being. If 2020 proves anything, these goals cannot be achieved alone. We all depend on each other in a million small ways and hopefully kindness blooms this year.

With pizza pies and tired eyes,


Source link

Be the first to comment

Leave a Reply

Your email address will not be published.