Scouting: looking for an excel sheet

[Jon Stratis]

Originally Posted by Captain_Kirch

If you want something easier I suggest using pivot tables. Tableau is pretty simple it would replace that second sheet you mentioned.

No, not really looking for something easier. Pivot tables are, frankly, a little too easy for what I'm looking for. Doing the macro's gives us a great excuse to get students involved with a second programming language (even if it is just VBA) and get them to think a little differently about programming, as the application of it and the language itself is way different from what we do for our robot! I wouldn't give up on the chance to give them that experience just to save us an hour or two of effort.

[TimTheGreat]

Originally Posted by Boltman

Personally would not go with "stats" they can be very misleading..

This is match statistics.

[evanperryg]

Originally Posted by jojoguy10

I'm really liking gear scout, but I can get it to save the data. It says that "team, match, regional, or your team number is missing." I was trying to test it with the Auburn Mountainview District. That regional code was WAAMV, right?

Go up to the options menu by hitting the menu button in the top right corner. In there, you need to set your team number, your name, the official FIRST event code for the event you're scouting, and your preferred save method. I'll talk to the software devs to see if we can make that toast (yes, the little pop-up things are actually ccalled toasts, even in the code) a little more clear. Best of luck, and I'm always happy to help you guys develop the best implementation of GearScout possible for your team!

[jojoguy10]

Originally Posted by evanperryg

Go up to the options menu by hitting the menu button in the top right corner. In there, you need to set your team number, your name, the official FIRST event code for the event you're scouting, and your preferred save method. I'll talk to the software devs to see if we can make that toast (yes, the little pop-up things are actually ccalled toasts, even in the code) a little more clear. Best of luck, and I'm always happy to help you guys develop the best implementation of GearScout possible for your team!

Right, thanks, I get that, but I wanted to make sure I understood what you guys are looking for as in "Regional Code". Can you give me an example (that your code is looking for)?

[evanperryg]

Originally Posted by Boltman

Personally would not go with "stats" they can be very misleading..eyes on bots is way to go.

Qualitative observation comes with inherent bias, and it's extremely noticeable from qualitative scouter to qualitative scouter. I've got one who's been doing it for 3 years now and I'm pretty sure half his notes say "don't pick" somewhere inside them, while I have other scouters that sympathize with a robot that hasn't moved for most of an event. (Though after Waterford, I guess the notion that a dead robot is a bad robot no longer applies) No type of data can be utilized as effectively as possible without the other types. Pit data, match stats, and qualitative observations are all important factors in making strong picks and making strong strategies.

I know bots better than even their drive team does...

You don't. You might think you do, but you don't. If 4 years and over 120 matches at the helm (like the medieval reference?) of the 2338 scouting team has told me anything, it's that scouters will never know everything an astute drive team will know. Even if they aren't willing to admit it, that drive team knows each and every thing that could possibly go wrong with their machine, each angle that, when pushed, will easily throw their robot out of position, exactly how tall and how close a blocker needs to be to shut them down. Everything. Details that could never conceivably be seen by a scouter, because they are details that couldn't be seen unless a one-in-a-million situation occurrs in which that weakness is exposed.

For instance stats said we had goal points when we did not fire in a single goal yesterday. Go figure.

Sounds like either those scouts are poorly trained, or they don't have an effective, ergonomic scouting system. After having problems with data accuracy in 2014, we switched to our current system and have found great success with it. If your data is clearly wrong, consider redesigning your entry method.

Originally Posted by jojoguy10

Right, thanks, I get that, but I wanted to make sure I understood what you guys are looking for as in "Regional Code". Can you give me an example (that your code is looking for)?

Go on TBA, and find the 4 or 5 character code next to the name of the event. For example, the Midwest Regional is "ILCH" and FiM Waterford was "MIWAT." I'll be honest, we don't do any data validation on the code you put in there, besides the character count. However, keep in mind that once we release the publicly-available support for database queries, you'll need to enter whatever code you used in order to get your data, so I'd strongly suggest making sure all of your scouters use the same code.

[logank013]

This year, our data is very simple. Simple data is the best data. We have 6 scouters and one per team. Our teleop data is as simple as this. How many high goals? How many low goals? What 4 defenses are on the field and how many times was each defense crossed from the neutral zone to the courtyard? Capture? Scale?

We have a macro that is super useful and was written in about an hour all while teaching me how to do it in the future. This macro works like this. We are able to type in what defenses are on the field and how many times each was crossed. The macro lets the 4 defenses that aren't on the field to be left blank. This causes the sheet to sort the data so that the averages are not messed up. We also have a macro that sets up 60 sheets in a matter of second. We have 2 sheets per team. We type in the team list and it sets up both sheets and organized them for each team.

We ran into some problems at this event between Chrome OS and Windows which caused us to do paper scouting. After entering all the data, everything worked as planned. We have lots of great tools like defense predictors. It uses our data and tells us which defense is the best to put out each match. We built this database in about 5 meetings (spread out). Our scouting programmer and I have already talked about how to fix our issues. By our next competition, our goal is to have live data that automatically pulls into the database with the push of a button. Our laptop issue should have a quick fix that will allow us to go paperless (the best type). All I'll have to do to set up the database is enter in the team list and schedule.

Databases don't need to be lots of work and don't need to be expensive. Most of our tools that were used the most at this last competition were developed on our 11 hour bus ride to the competition. They are simple to make but are still powerful when looking for teams as a captain. If you are interested in how we use excel to create our database and have questions about how we use it, feel free to PM me. Thanks.

[Boltman]

Originally Posted by evanperryg

Qualitative observation comes with inherent bias, and it's extremely noticeable from qualitative scouter to qualitative scouter. I've got one who's been doing it for 3 years now and I'm pretty sure half his notes say "don't pick" somewhere inside them, while I have other scouters that sympathize with a robot that hasn't moved for most of an event. (Though after Waterford, I guess the notion that a dead robot is a bad robot no longer applies) No type of data can be utilized as effectively as possible without the other types. Pit data, match stats, and qualitative observations are all important factors in making strong picks and making strong strategies.

You don't. You might think you do, but you don't. If 4 years and over 120 matches at the helm (like the medieval reference?) of the 2338 scouting team has told me anything, it's that scouters will never know everything an astute drive team will know. Even if they aren't willing to admit it, that drive team knows each and every thing that could possibly go wrong with their machine, each angle that, when pushed, will easily throw their robot out of position, exactly how tall and how close a blocker needs to be to shut them down. Everything. Details that could never conceivably be seen by a scouter, because they are details that couldn't be seen unless a one-in-a-million situation occurrs in which that weakness is exposed.

Sounds like either those scouts are poorly trained, or they don't have an effective, ergonomic scouting system. After having problems with data accuracy in 2014, we switched to our current system and have found great success with it. If your data is clearly wrong, consider redesigning your entry method.

Go on TBA, and find the 4 or 5 character code next to the name of the event. For example, the Midwest Regional is "ILCH" and FiM Waterford was "MIWAT." I'll be honest, we don't do any data validation on the code you put in there, besides the character count. However, keep in mind that once we release the publicly-available support for database queries, you'll need to enter whatever code you used in order to get your data, so I'd strongly suggest making sure all of your scouters use the same code.

Cool thanks for the tips.

[bennettj800]

I have one for you!!! yay!!!


I have spent sleepless nights and homework hours working on this sheet, so i really hope you like it...

|Click Here|
if that didnt embed:
https://drive.google.com/open?id=0B4...FpFTC04d2doYjg
please

please use it.

also if you do, i suggest you use the sheets provided, since it will be the collection system.

pm or email me if you have any questions (my username is bennettj800 )

thanks for posting this thread! i have been wanting to upload this for a while now

[choltz]

It sounds like quite a few teams rely on what I would call qualitative scouting rather than quantitative data. IMO qualitative will lose every time to quantitative because qualitative is subjective at best and tends to reflect the scouters' beliefs of a team based on: past years, team size, rookie status, robot look etc.

Similar to the plot of moneyball, if your numbers accurately measure what you want to know, say percentage high goal accuracy, then they will be much more accurate at finding trends in the 60+ teams you have to simultaneously watch playing for two days straight. At least compared to the opinions of a small group of experienced team members.

[bennettj800]

Originally Posted by choltz

It sounds like quite a few teams rely on what I would call qualitative scouting rather than quantitative data. IMO qualitative will lose every time to quantitative because qualitative is subjective at best and tends to reflect the scouters' beliefs of a team based on: past years, team size, rookie status, robot look etc.

Similar to the plot of moneyball, if your numbers accurately measure what you want to know, say percentage high goal accuracy, then they will be much more accurate at finding trends in the 60+ teams you have to simultaneously watch playing for two days straight. At least compared to the opinions of a small group of experienced team members.

While i totally agree with what you have said, those things dont totally correlate with moneyball.... although i see the point you make here- we chose who we chose regardless, and no data is gonna get in our stinkin way!

the thing is, this spreadsheet collects everyone's information in one place and tidily color codes and organizes it so that we can percieve each team more objectively, and much less subjectively. we can see whta they claim during pit scouting in the case that a) they dont get to show off or b) they are poorly reflected on the field. this helps us with strategy.

My hope in scoutmaster this year was to eliminate all hope for the subjective chosers and look at things like a machine, to eliminate names and reputations and go for the numbers on their own.

of course, sample size is affected when teams play 7-10 matches at most, but... we cant change that...

i hope someone uses this before its too late!