Tableau at PNW Districts

[Sunnykx]

Scouting Recap of Glacier Peak.docx

Our Team (CPR - #3663) used Tableau to analyze our match scouting data at the Week 2 PNW Glacier Peak Competition. I am hoping that I successfully attached a copy of our Tableau dashboard in this post. Tableau has been a huge factor in our scouting the last two seasons. We were introduced to it by Quinn Schiller on Team 1983, Skunkworks Robotics. It helped us become a captain in our division at World's in 2015 (losing to 1114 and 148 in the Finals) and contributed heavily to our winning Glacier Peak last weekend.

Tableau is a very powerful tool. It takes our dry Excel data and transforms it into beautiful pictures and graphs that help organize and clarify our information. There is a bit of learning curve with it, but their are multiple tutorials on the Tableau site and through FIRST that can help answer any questions.

At Glacier Peak, we used our Tableau dashboard, hopefully attached here, to convince the top seed, Team 2522, to pick us in alliance selections. Prior to our lobbying, they weren't even considering us. We showed them that we were a perfect complement to their team (we are heavy shooters and breachers, they are heavy breachers and they shoot some low goals). Having been alerted to our existence by our compelling data, they then watched us in our next match. We made five high goals and breached heavily, including the Category A defenses (Portcullis and Cheval de Friese). They then watched the number two seed, who had similar data, perform. We drove more aggressively and quickly and adjusted on the fly to our partner's difficulties. They were convinced we were the best choice and picked us. We went on to win the competition.

This brings me to the subject of qualitative scouting. We have eight people dedicated to quantitative scouting (6 match scouters, 1 Excel data entry person, and 1 Tableau analyst). We have two scouters dedicated to qualitative scouting. We use the quantitative data as the basis for all match strategy preparation and then incorporate the qualitative data to make final decisions. The qualitative data helps us evaluate driver skill (critically important in this year's competition), robot speed and consistency, the flexibility of a team to change strategy mid-match if necessary, and how much partners work well together. Quantitative data can suggest this information (consistency in scoring across matches, amount of defenses crossed and points made to suggest speed) but simple words such as "unreliable" and "fast" convey a wealth of information quickly. The qualitative and quantitative data together help provide a complete picture of a team and provide checks and balances for one another.

I am hoping that more teams will use Tableau. I love it when I see teams making informed, wise decisions in match strategy and alliance selections. I think it elevates the level of play for everyone. While not everyone has the staffing available to make up a full scouting team, it's not hard to combine with other teams to pool data. Our team is always willing to work with other teams to share our process and we post our bar charts at competition.

I'd be curious to learn if other teams use Tableau or something similar.

Good luck to everyone at Stronghold!

[Ed Law]

This is the third year 2834 is using Tableau. We like to collect data and we like them shown in a clear and concise way to make good decisions. I can't give all the credit of how we did in the last 3 years all to Tableau but it certainly is one of the enablers.

[SoftwareBug2.0]

A word of caution: Tableau is not a cure-all. This should go without saying, but you have to know both how to use it well and understand what you ultimately want to get out of it.

One year we had a new mentor who decided he was going to do analysis of our scouting data using Tableau and he made a bunch of interesting graphs. They were interesting ways to look at the data - he was not a dumb guy. Unfortunately, they didn't quite line up with what we wanted to know, and we weren't able to get the data we did want out in time so we ended up totally disregarding all of our scouts' work at an event. This was a couple of years ago. It was unfortunate.

I know that teams have had good luck with it, but be sure that you know what you're doing before you jump in without a backup.

[Sunnykx]

I agree. We use qualitative data as a balance check for the Tableau data. Also, as in any data collection scheme, you need to know which variables to scout.

[Ed Law]

I agree 100%. During the build season, the scouting team meet and propose what scouting information to collect. It is then reviewed by the whole team. After it is finalized, programmers update the Android scouting app and the Tableau person prepare the graphs and work with the game strategy team to finalize the graphs.

The other nice thing about Tableau is once you have the data collected, you can look at it other ways on the fly relatively quickly if the graphs you prepared ahead of time is not as useful as you expected.

It works for us because it is a team effort. I hope more teams will try the software. It is great.

[Sean4488]

We've also begun to use Tableau this year with data from our scouting app. It's been hugely helpful not just for alliance selection, but also pre-match strategy, and convincing other of our strategical ideas.
Really helpful tool, and available to all FRC teams, it's really a shame people don't take more advantage of it.
It's really great to hear that other PNW teams are using it, I've been curious about that for a while.

[jkoritzinsky]

Quick question for you guys: One of the possible features for SuperScouter for next year is the ability to export a Tableau Data Extract file with the table generated from the schema. Would any of you find that interesting?

[Sunnykx]

Originally Posted by Sean4488

We've also begun to use Tableau this year with data from our scouting app. It's been hugely helpful not just for alliance selection, but also pre-match strategy, and convincing other of our strategical ideas.
Really helpful tool, and available to all FRC teams, it's really a shame people don't take more advantage of it.
It's really great to hear that other PNW teams are using it, I've been curious about that for a while.

I'm really excited that Shockwave is using Tableau! We're pretty sure we've qualified for PNW District Champs so we'll see you there. I'd love for our Tableau student analyst to meet with yours and talk about Tableau.

[Sunnykx]

Originally Posted by Ed Law

I agree 100%. During the build season, the scouting team meet and propose what scouting information to collect. It is then reviewed by the whole team. After it is finalized, programmers update the Android scouting app and the Tableau person prepare the graphs and work with the game strategy team to finalize the graphs.

The other nice thing about Tableau is once you have the data collected, you can look at it other ways on the fly relatively quickly if the graphs you prepared ahead of time is not as useful as you expected.

It works for us because it is a team effort. I hope more teams will try the software. It is great.

Our team met during build season, too. Our software team was able to make a sample dummy data set of 24 teams with different Stronghold abilities. I had my scouts use Tableau data to strategize for different red and blue alliances like we do in a real competition. They were able to get used to how the data looked in Tableau and we were able to make changes to the categories of information we looked at. We also did sample alliance selections based on the data. Given that we were all pretty familiar with how this year's competition data looked before we even had a real competition, we were able to hit the ground running. We have had to make very few changes to our data collection after two competitions. Again, I will say that qualitative data is a necessary balance to the quantitative data.

[Navid Shafa]

Originally Posted by Sunnykx

We were introduced to it by Quinn Schiller on Team 1983, Skunkworks Robotics.

It's thanks to this former student that Tableau now provides a copy for every team in the Virtual KOP. It's great to see the legacy he left in PNW and beyond.

[Sean4488]

Originally Posted by Sunnykx

I'm really excited that Shockwave is using Tableau! We're pretty sure we've qualified for PNW District Champs so we'll see you there. I'd love for our Tableau student analyst to meet with yours and talk about Tableau.

Absolutely, I'd love to meet your analyst, we have a few people using it, but primarily me, so I'd be excited to meet others. See you guys at Dist. Champs!

[alextried]

Originally Posted by jkoritzinsky

Quick question for you guys: One of the possible features for SuperScouter for next year is the ability to export a Tableau Data Extract file with the table generated from the schema. Would any of you find that interesting?

Yes please! What we're planning on doing right now is SuperScouter -> Excel -> Tableau.

[Sunnykx]

Originally Posted by Sean4488

Absolutely, I'd love to meet your analyst, we have a few people using it, but primarily me, so I'd be excited to meet others. See you guys at Dist. Champs!

Awesome!

[Alex Webber]

Is there a product key in the KOP? I might ask my leas mentor about downloading it on our Monday meeting right before the IA regional.

I've heard it's super awesome, and with our awesome scouting system, a printer in the pit, and this, that could round off our data. Not to mention look pretty.

[*Rachelle*]

We've considered using Tableau as well, it's actually on my laptop, I just haven't dedicated the time to learning it proficiently enough for competition use.

Not to mention we do not have enough students to have one specifically set on data input, I did it for two years and found that we could analyze the handwritten sheets much more effectively. It is mostly a tally system with a section for driver skill, so while we may not have the graphs, it is still quantitative.

My first question would be how long do your scouting night meetings generally take with the help of Tableau? Oftentimes when teams are talking about how they scout most efficiently I can't tell if efficient to them is 1, 3, or even 6 hours of hashing out the data.

Our current system in the last two years has been great for us, I'd be interested in seeing what your match scouting sheets look like as well. Would you mind sharing?

Congratulations on the win in Glacier Peak, those finals matches were fun to watch!