Visually viewing Caleb Sykes' Scouting Database: Data is Beautiful

Like last year, I’ve taken the awesome data available in Caleb Sykes’ Scouting Database to present the data visually. Teams are plotted with one another in order for the viewer to get a feel for how a team’s performance ranked among its peers. Plus, the location of a team’s “bubble” tells a good story about the overall performance of the robot in 2019 so far. Don’t forgot to pause over a team to get all the stats via the pop up.

There are a lot of teams so you’ll probably want to filter down to your district and/or event (only historical events are present). You can also highlight a team number using the Highlight Team filter.

View for Game Element Analysis

View for Overall Points Earned Analysis

As Caleb updates his database in the coming weeks, I’ll be updating these reports as well.

Please let me know if you have any questions.

Hope this helps in your team’s preparations for the second half of competition season. Good luck and HAVE FUN!

Remember… Data is Beautiful!

Cheers,
Alicia Bay
1918 NC GEARS

37 Likes

Hmm, it looks like you have the rookies in the FMA district assigned to “fma” and the veterans assigned to “mar”. I don’t know where you’re getting your data from but it’s probably trash.

11 Likes

Nice Caleb. :smile:

If “FMA” and “MAR” are supposed to be one district, you can always filter on both of those “districts” to see all the teams together.

This is the workaround until the Data Master gets his districts cleaned up in the source data, LOL.

This is awesome! Thanks for doing the work for the benefit of all, I was just wondering how average game piece scoring is determined. I assume this is taken from tba. Do you just divide the total number of game pieces scored by an alliance by the three robots and give each member an equal share of the points?

Is there a reason why there are teams with negative points? Is this calculated similarly to OPR, so they account for fewer points scored by their partners?

2 Likes

This is great, thanks!

Where is the source data for these graphs?

Caleb Sykes’ Scouting Database posted here on GitHub, and he pulls from TBA.

Wow thanks for this it’s really nice to see this data in a way that makes sense!

You are most welcome. Data is Beautiful!

1 Like

Awesome work Alicia. Thanks for putting this together.

Any chance for MSC you might be able to have it sorted by division assignments so we can view just the 40 teams in a division?

1 Like

Hey Nick, great question. Once we are a little further into our season (probably around week 6), I’ll add some more bells ‘n’ whistles that will enhance pre-scouting for both MSC and Nationals. Stay tuned!

Hi Alicia! Hope all is well with you! How about forward looking? For instance, we’d like to look at East Kentwood and figure out how to beat 1918 :slight_smile:

1 Like

Great stuff. Thank you. I would love to see actual individual team “scouted” data vs the TBA team calculated data on a graph.

HEY MATT WHITE! Ahhh, the crystal ball view via historical data… still working on that one, Matt. I may have to pull on R or SAS to do some predictive modeling, although my sample size is pretty small. We need a longer qual season to suffice us data nerds. :sunglasses:

1 Like

No problem! Maybe allowing a robot filter so one could select the robots they want to compare (using whatever data you have at the time)? I am sure you have nothing better to do! :wink:

Awesome graphs! We will probably use these to gauge our own performance as well as use them to see what we’re up against at upcoming events. Will you be updating these each week?

The graphs are beautiful…

I am looking at some of the data from the event we were at last weekend and there are some weird data points. I am talking about robots that didn’t have cargo mechanisms with cargo scores >6.5. I get that placing hatches can lead to other alliance members doing better on cargo, but I don’t think that is what is going on here.

I am wondering if underlying data isn’t capturing a noticeable “how often their alliance saw defense” component.

For example:
Lets say team A can run 4 hatches and 8 cargo undefended but runs 2 hatches and 4 cargo while defended. The result is the teams with team A while team A is defended will have their numbers pulled lower. The teams with team A while team A is undefended will have their numbers pulled higher.

If there is a team B at the event that happens to be on a bunch of alliances that see very little defense, their ratings are going to be much better than they are.

2 Likes

Actually, I was quite pleased at how closely my team’s scouted data from our first event (FIM Lakeview) and the TBA data matched. Both charts are published on the 1918 Scouting Viz Page located here. You’ll need to set the filter on the bubble chart to FIM Lakeview.

Best to view both charts side-by-side in two different browser windows.

The chart types are completely different but tell a similar story.

1 Like

You are correct. Data coming from TBA, while awesome because it covers a lot of teams at once, can only go so far. This supports the fact that a team’s own scouting data (both pre-scouting online matches and live scouting at the event) to be #1 above all else, undiluted and spot on for what your team is looking for. This bubble chart is another tool in your scouting toolbox to aid your team in preparing for your next event. Data is still Beautiful! :sunrise_over_mountains:

1 Like