Challenge #100: Find Ned!

Happy New Year, and welcome back to Weekly Challenge! Here's to another year of stretching our skills, exploring new tools and encouraging each other to be the best Alteryx-ers we can be!

Today is a day that so many of you have awaited: Challenge #100! Not only do you earn bragging rights for completing 100 Challenges but this Challenge also gets you another shiny badge to add to your collection! We have 8 Challengers who are neck and neck with 99 completed Challenges, with a couple more contenders not too far behind. Check out the updated Weekly Challenge Index to see the current standings. Ready to get started? We're off to the races!

After a very busy 2017, @Ned really needed to get away from it all. So, he did what any of us would do: he set up his tent in the farthest location from any road in the state of Colorado. But, now it’s 2018 and we’re all ready to get back to work on completing the newest course of Interactive Lessons, passing the Advanced Designer Certification exam and preparing for Inspire 2018. All of us…except for Ned. Where exactly is Ned? And how quickly can we find him? Well, that's what you need to figure out!

You are provided with two Input datasets: ColoradoRoads.yxdb, which contains spatial data of all the roads in Colorado, and Colorado.yxdb, the administrative boundary of the state. Using these two files, you need to not only find the location of Ned's tent (within 100 meters of the correct answer) but also locate him the fastest you possibly can. You'll notice that the start file for this Challenge does not include the solution! To check the accuracy of your solution, plug your final coordinates into the app called "HowCloseIsNed", which is attached to this post. Not within 100 meters of Ned's tent? Can’t find him within a few minutes? Optimize your process, and try again! Share your results for distance and run time when you post your solution.

Due to file size limitations we can not attach the files to the post,please download by selecting: challenge_100_start_file.yxzp and HowCloseIsNed.yxwz

challenge_100_solution.yxmd

Spatial Analysis

Preparation

Advanced

Data Analysis

Transform

Spatial

Difficult

Accepted answers

All comments

Ned

I'd help people, but there is no internet access all the way out here... ;-)

patrick_digan

My method isn't fool proof, but I got within 100 meters of @Ned (even if it happened to be a happy accident!)

Spoiler

I just set up a really big grid, used the find nearest against the roads, and then chose an arbitrary cutoff of 8 km which got me down to 5 grids. I further split those grids down to 100 meters and ran the find nearest again. The winning grid had a centroid only 17m from Ned! It took exactly 1 minute to run.

challenge_100_start_file.yxmd

ErikaW

Reply from Ned received via carrier pigeon.

2018-01-08 at 1.25 PM.png

Natasha

@patrick_digan oh wow, that was quick!

Spoiler

I feel that I am going down the rabbit hole trying to work out yxwv, batch and run the app

chris_love

When @jdunkerley79 says its hard then you look up from your work and pay attention. Took a little while to work out the best method, but I left it and came back for a second go before bed. Runs in 2mins 4 seconds.

Spoiler

I won't give away my full method but I used a grid to work outwards from the roads (I wanted to use buffer but it took an age on my PC, I'm sure it works with polylines), then inverted them against the state polygon to create islands of likely locations. Then I narrowed down the options by building two sets of grids, one at 1km and one at 0.1km narrowing the options after the first.

DataPrepChad

This is eerily reminiscent of a Grand Prix challenge around 7 years ago...it stumped me then but did not this time. My point was just under 22 meters from Ned.

Runs in around 2 minutes.

Spoiler

challenge_100_start_file_ChadMartin.yxmd

LordNeilLord

Congrats to @patrick_digan for being the first to get to 100 challenges!

I'm not sure how I managed to get an answer as towards the end I was just throwing tools onto the canvas and hoping for the best...anyway I'm done, it's late and I'm going to bed.

Spoiler

Same principle as Patrick & Chris...

-I chopped the roads up into a number of grids and then looked for the most "empty" one.
-I then cut that one up into smaller grids and looked for a grid that was the furthest away from a road

I managed to get within 55 meters and it took 1:01 to run

Weekly Challenge 100.png

challenge_100_LNL.yxmd

philipmannering

Great challenge. There are few that get you to have to think about optimising.

My solution finds Ned to within 34.5 metres in 46.5 seconds.

Spoiler

challenge_100 - Finding Ned.yxmd

David-Carnes

Went through several iterations, but...

Congrats, you got within 10.2 meters of Ned. Look to the S

Spoiler

David-Carnes - 100.yxmd

darryl5280

Dropbox is blocked at my company as other companies like mine source code should never be stored on external sites. Can you place the data files on https://github.com/alteryx as GitHub is not blocked for me?

jdunkerley79

Finally got mine tidied up and working... pleasantly close to Ned (0.4 m) run time 1m 16s

Spoiler

Congrats, you got within 0.4 meters of Ned. Look to the NE

challenge_100_jd.yxmd

Ollie_Power

Got to within 17.1 meters of Ned in 14 seconds

Spoiler

challenge_100_solution_OlliePower.yxmd

ggruccio

Since the input files are large I only included the workflow. Found him within 2.2 meters!

Ran in 5:59 on my computer, but the weekly virus scan is running at the same time so it may have been competing for resources.

challenge_100_gg_finish.yxmd

ChristineB

@darryl5280, I'll see what we can do! I'll post back when I have an update!

CarolineMurnane

Spoiler

I considered another iteration, but 0.3m in 2 minutes will do!

Ilías

Okay, I've never done spatial analysis before so this challenge was quite difficult for me. After hours of hopelessly wondering in the Colorado outback I decided to get some help and was inspired by @LordNeilLord 's solution. Anyways, learned a lot about the spatial tools and think this was a fun excercise!

Challenge 100 ili.yxmd

Ned

So I think we need to organize an expedition to this spot next summer. It looks like its about 16 miles hiking from the nearest trail head (one way), so its probably at least 2 days round trip. I think I've found the best route:

https://goo.gl/maps/H2VEPtzAUYU2

Who's in?

Ned

Or maybe I found a better route. Hard to measure distance because google doesn't have the trails.

Starting from the south: https://www.hikingproject.com/trail/7032122/pine-river-trail-523

Probably a similar distance, but less vertical.

Spoiler

estherb47

This one was a fun exploration of the spatial tools. After many different attempts and approaches, landed on using grids of shrinking sizes. Rinse and repeat the gridding process, finding the one furthest from the roads.

I wanted to embed the "How Close is Ned" app into my workflow, and create an iterative macro that would loop until the workflow gets you to under 1 km away, but couldn't figure out 1) how to call the app, and 2) how to get the results written to a format that can be used for comparison. Would love suggestions & feedback here!

25.1 seconds is very reasonable to get within 3 m.

Spoiler

Workflow:

challenge_100-EHB_solution.yxmd

LordNeilLord

@estherb47 0.3 meters! That's got to be some sort of witch craft

NicoleJohnson

Hoorah for Challenge 100! And to finding @Ned (0.3m away in 21.5 seconds)!! Looks like a great spot. Maybe we should host Inspire 2019 out there in the wild Colorado yonder...

Spoiler

Ended up with basically the same solution as several others... I spent a LOT of time playing around with buffers, spatial process, and varying orders of tools, but this definitely seemed to be the fastest/closest method.

Super fun challenge #100 @ChristineB!! It's been lovely reaching these last 100 milestones with you, @JoeM, and the rest of the Community. The view is quite lovely up here!

Cheers,

challenge_100_NicoleJohnson.yxmd

Natasha

I finally got my head around this challenge which was quite challenging. Final results are 3.6m from Ned in 22.2 seconds

Spoiler

At first I went completely the wrong way of thinking about this. Since I was not sure what I need to look for, I imagined this to be some sort of iterative process of checking the answers and I was looking for a way to automatically call an app from run command tool and populate it with app value files. Obviously I failed.

After chatting to @PhilipMannering I realised that I missed a very crucial information while skimming through the task, and that is Ned set up his tent in the farthest location from any road in the state of Colorado. Well now all the grid solutions started to make sense to me. I spent some time to figure out what is the correct relationship between the target and universe, and went on to decrease the size of the grid. At the end I think my solution is quite similar to @estherb47.

Screen Shot 2018-02-11 at 20.42.09.png

challenge_100_Natasha.yxmd

JosephSerpis

Challenge Completed took over a minute

Challenge_100_Joe_Serpis.yxmd

iain_sterland

0.9 metres in 29.4 seconds

challenge_100_IainS.yxmd

jamielaird

71.8 meters in 20.2 seconds

Spoiler

challenge_100_JL.yxmd

SeanAdams

So - took a week and a half to go through this - tried some pretty silly ideas on the way.

Solution details in the spoiler section - around 2m30sec (if you exclude the weeks it took me to get a solution that ran in a sensible time)

Spoiler

The solution that I got to just uses progressively larger buffers (around the roads) in an interative macro to find the value of a buffer that eliminates all but the final spot where Ned is hiding.

This is the main flow - it first puts the roads into a buffered polygon, and then looks for ever larger buffersput buffer around the roads. Broken into batches of 50 to speed this alongThen look iteratively for the largest buffer that still returns one hiding place
The solution posed by @LordNeilLord ; @JosephSerpis ; @Natasha @ggruccio @estherb47 @DataPrepChad; @jamielaird @NicoleJohnson; @patrick_digan using a grid then a spatial match and/or filter for nearest is much more effective.

challenge_100_solution_SeanAdams_v4.yxmd

ConsolidatedRoadMaker.yxmc

SpotFinder.yxmc

jasperlch

challenge_100_JL.yxmd

kat

This must have been the hardest challenge I've ever done. Took quite a while to figure out and the final product just seems so inelegant. Nonetheless, just under 20 seconds to run and 33m away.

Spoiler

Challenge #100.yxmd

StevenBetts

Here's my solution

Took me longer to get my head around this than I'd have liked, but once I figured out the principle then it went pretty smoothly.

72 metres away. Workflow runs in 16.1 seconds.

Spoiler

challenge_100_start_file_SMB.yxmd

catherineduffy

Within 5.5m in 26.9 seconds.

challenge_100_catherineduffy.yxmd

JoBen

Spoiler

Cheers! That was a lot of fun!

challenge_100_Joben_Solution.yxmd

dsmdavid

I first did it as large grid, then medium grid, then fine grid but, after seeing so many posts with the same approach, I tried using iterative macros to find a locale max after the first grid. Still a decent job at < 25 s.