Advent of Code 2024 Day 12 (BaseA Style)

AlteryxCommunityTeam

Discussion thread for day 12 of the Advent of Code - https://adventofcode.com/2024/day/12

Advent of Code

Accepted answers

All comments

AncientPandaman

take me awhile to remember how to use "Make Group" tool. save time from build manualy.

thanks for past events got similar questions.

Spoiler

part1:
1. get adjacent cells, if neighbour not same, then it counted as perimeter.
2. get the region id.
3. join and make the group by and calculation etc

part2:
scan for side by side, use multiple rows to identify the sideID

eg:

xxx<--1
xx<--2
xxx<--3
xxx<--3

if [row-1:y] = y and [row-1:x] = [x]-1 then [sideID] else [sideID] + 1 endif
Screenshot 2024-12-12 151657.png

another tip:

Spoiler

remember to check the sorting, i.e. check for engine compatibility mode for AMP.

Day 12.yxzp

Gawa

I applied spatial analysis to both part 1 & 2. It was hard to solve for me.

Spoiler

Part1:
1) Define 1 x 1 square for all tiles, and create regions + outer frame object(surrounding the entire map).
2) Buffering each regions a bit and matching each other will create perimeter objects. Calc their length and area of regions by Spatial Info tool. *Length will be double of the length, so divided by 2.
3) Cal unit area and length of 1 x 1 square separately, and divide the result of 2) by these unit values.

Part2:
Using Multi Row Formula, count how many perimeters exist for each region. To be honest, the logic I applied was finalized through try and error, and it contains fuzzy logics.

2024D12_gawa_rev1.yxzp

CoG

Very cool problem with application in (ML) Image Processing for edge detection.

Spoiler

Part 2 can be solved by understanding that if you shift all cells in each of the cardinal directions, then remove all cells that overlap within the same group, all you are left with is a side/edge. Count all such sides, and multiply by the number of plots in a group (area) and sum to achieve the answer:

V2 (no macro; thanks to the very powerful Make Group Tool + a lot of simplification throughout the workflow):

Spoiler

Main.yxzp

Main V2.yxmd

DaisukeTsuchiya

I did not use macros. For P2, I struggled by drawing diagrams in Excel.

Spoiler

For P1, I grouped adjacent elements using the "Make Group" Tool. A single plant would have fences on all four sides, but if it is adjacent to others, the shared sides are subtracted in the calculation.

For P2, I took the approach that irregularities (convexities or concavities) add more edges, and I calculated the edges by dividing the analysis into vertical and horizontal directions.

Spoiler

DT2024_Day12_Rev0.yxzp

Yoshiro_Fujimori

I got stuck with Part 2 and got a hint from @gawa that Spatial Tools works, and it actually worked. Thanks gawa-san!

Spatial Tools are very powerful in "visualizing" the process.👍

Spoiler

Part1
Used Make Group Tool to find the areas and counted the neighbors of each plant.

Part2
Create squares for each plant and combine them with the area ID.
Split it to Regions with Poly-Split, Convert it to boundary with Formula,
Split each to Points, and then count the corners referring to the coordinates with Multi-Row Formula.

Day12_workflow.yxmd

Spainey

I hadn't come across the Make Group tool before but it did exactly what I needed for this challenge when I was struggling to identify the area and perimeter of distinct groups. I didn't use any macros as I was able to find and check the values of adjacent cells using Generate Rows. I did find Part 2 very tricky and was originally considering a convoluted method to try and count the number of corners (because corners = edges), but couldn't wrap my head around the best way to do that. In the end... (see spoiler)

Spoiler

For Part 1, I considered the adjacent cells that matched the current cell, which allowed me to create groups using the Make Group tool.

For Part 2, however, I considered the adjacent cells that did not match the current cell, which, with a few headaches figuring out the correct sort order and formulae, enabled me to check if the previous unmatched cell was part of the same edge or part of another edge - based on row/column sorting. By finding the greatest count of edges per direction and then aggregating that figure on a Group level I could join to the Area data calculated in Part 1 and calculate the discounted price.

AoC 2024 Day 12.png

2024_AoC_mark-spain_Day_12.yxmd

AkimasaKajitani

Solved!

Spoiler

I got to the Make Group tool in Part 1 early on, but took a long detour in Part 2. When I first tried Spatial tools in part2, I was disappointed by the extra points that were left over. I then tried to use an Iterative macro to identify the edge, but it didn't solve some particular patterns. I ended up using spatial tools to identify the order of the perimeter points, and then eliminated the straight line ones. I used some of the logic from the detour, so maybe it wasn't a waste in the end. I solved it just before going to bed somehow.
I remembered too late that polygons basically hold their points in a clockwise order.

Day12.yxzp

leozhang2work

Spoiler

Taking hint from @gawa's spatial approach, part 1 become very easy

day 12.yxmd

Hub119

Fun use case for spatial tools today 😀

Spoiler

AoC 2024 Day 12.yxzp

ntakeda

My solution.

Spoiler

Mark records for which record have already found a group.
When all record have been marked, finish.
It takes 4 minutes because it is not running in parallel...

AoC2024-part12.yxzp

DataNath

Finally finished up with #12! Spent an insane amount of time going down rabbit holes that would all work for all but one of the sample grids, before taking a few hints from here & getting over the line!

Spoiler

mmontgomery

Day12. More in spoiler

Spoiler

P1: Make group was the key piece to tie elements together then I created a new field called new_data to organize it then find boundaries
P2: I struggled for a long time and had a similar approach to @ntakeda and his final part inspired mine to finish it

Day 12_MM_Final.yxzp

DaisukeTsuchiya

I tried solving Day 12 P1 and P2 again using the Spatial Tool although I solved it without spartial tool.
Last time, I struggled a lot, drawing diagrams in Excel and pondering over them, but this time, I was able to solve it very easily.
As expected, the Spatial Tool at bottom container is much better because it's more visually intuitive.

Spoiler

DT2024_Day12_Rev1.yxmd

geoff_zath

Spoiler

AoC_2024_day12_v3.yxzp

Tokimatsu

Solved without macro.

Spoiler

AoC2024_Day12.yxzp

Samantha_Jayne

And she is finally back in the game. Thanks to Cory on the WhatsApp for helping me find what I was missing. Turns out my dataset also had a few red herrings in there too. But I finally found all those. Phew. Spoiler below. #darkmoding

Spoiler

Samantha_Clifton_Day12.yxmd

kelsey_kincaid

This one took me awhile to wrap my head around. Part 1 was one thing, Part 2 was another. After looking at some other solutions, @Yoshiro_Fujimori 's really resonated. Thanks for your insights!

Spoiler

Part 1: Took some time to find the right approach, but the key is just to determine how many adjacent cells do not have matching data.
Part 2: I started by trying to determine which cells were corners, but without leveraging spatial tools. It's doable, but I was having a hard time visualizing and wrapping my head around the logic. Converting to spatial to get a visual of the corners and remove some of the noise from irrelevant cells was a game changer. I also learned a new formula - ST_Boundary - which I had never used before. That's what I love about Advent of Code - I ALWAYS learn something new!

2024_Day12_KelseyKincaid.yxzp

estherb47

Took the spatial route. Didn't do anything different than others here.

Spoiler

One thing I love about a spatial approach is being able to divide up the different shapes that have a common letter, using the Poly-Split tool. It's an easier way for my brain than make groups (which also works really well)

Also, fun finding by just messing around with the spatial tools. In part 1, if you use bounding boxes, the number of parts added up for each shape gives you the perimeter.

For part 2, calculated the point to point movement around the bounding boxes (a few poly splits take care of only catching the perimeter boxes of the shapes), and with a change in direction, increase the number of sides of the polygon. Unsure how a few of these wound up with odd numbers, which was throwing the total off, so if the number of sides were odd, subtracted one.

Day12.yxmd

ccanterman

A HUGE thank you to @DaisukeTsuchiya . I was trying to solve Part 1 the way that you did but instead of using the Make Groups, I just made a huge mess of it multiple times. I was actually pretty close with my first attempt, which was a macro to make the groups (only off by 2k for the final answer). When I used yours as a guide for doing a proper Make Groups and then seeing how to Join everything back together, I was able to solve it.

Part 2 was actually relatively easy, compared to Part 1. Excel helped me visualize it and then it was just a case of making sure my Sorts were in the right order

Please no one judge me for anything in the disabled containers. I'm so glad this one is done!

Spoiler

Part 1:

Part 2:

Day 12 v1.yxzp

lwolfie

Took a little bit. Almost skipped for later but I was so close.

Day12.yxmd

Quick Links

This months top contributors

atcodedog05 19598

Qiu 15867

binu_acs 15708

MarqueeCrew 13708

apathetichell 13703