Challenge #146: SANTALYTICS 2018 - Part 2

[OldDogNewTricks]

Attached is my solution.  I used a Python tool to do the website scraping (because why not?).  I navigate the pages if there are more than one (the provided scraping macro will miss data if the page count goes over 10) and I only add the valid solution types to the resulting data frame.  I then get down to a unique record per user in order to not double count solution postings.

For the monies distribution, I went with quantity over quality (as the 4+ star criteria is already applied) as well as the largest variety of items.  I also noticed that there were a few records with both the salesprice and msrp as null, I excluded those because I didn't figure anything was really free.

Spoiler

Result summary (before my posting):                             Result summary (after my posting):


Result detail (before my posting):

[Treyson]

I thoroughly enjoyed this one. I went through @patrick_digan's idea of circling back with accumulated left over spend so I implemented that. I also split clothes up into boys and girls in order to make sure there was even spend there as well. 

I also tried to make a pretty report but I wasn't able to get a report that actually rendered to match what I created in the "visual layout" tool in the laboratory. 

Anyway. Here's Wonderwall

Spoiler

[Paul-Evans]

Here's my results:

 

Spoiler

For the solution count, I used an iterative macro to go through each page of responses and create a list of all .yxmd and .yxzp attachments. 

To select items to purchase, I
- randomly sorted the items after filtering for the given criteria,
- selected items by type until hitting the spend limit, and 
- matched the remaining spending amount to the item with the price closest to that amount.

[Kenda]

Spoiler

I wasn't sure if the 50/50 split meant number of items per category or amount spent in each category. I went with the amount spent per category. Before my post, there was $200 available to spend. I started with the given macro but modified it a bit (in pink).

[mmenth]

I like how open-ended this challenge was!

Spoiler

For my solution I was given $400 to work with, so I first split that up so that we could spend $200 on clothing and $200 on hygiene. Since there were so few clothing products after filtering out products without enough reviews, with reviews below 4, and unavailable or limited products, I didn't add any filters of my own. However, there were a lot of hygiene products to choose from after applying the filters mentioned above, so I decided to look up the the products that A Precious Child accepts on their website. I found a short list and added that to my workflow. A quick find and replace helped me filter down the original hygiene list to just those products that the nonprofit would accept. I then sorted by customer rating and took the top unique products based on the their acceptable product list.

For my approach I assumed variety is preferred, so for the two product categories I duplicated my narrowed-down lists 10 times each, sorted by the rowcount, and calculated a running total. I kept adding products to my shopping list until the running total met my spending limit. If I had a remainder left over, I grabbed another product that would minimize the overall remainder for that category grouping.

Overall my list has 28 unique products, with 71 products total!

[ggruccio]

My approach goes after the higher ticket items...so maybe could help one specific family.

What a great cause!

[egranberry]

This is a great idea! Happy holidays! 

[aristiguieta]

I do many annotations (in the tool tags and as comments), so this may be helpful if you didn't know how to approach it.

I hope we get to the $2,000! 

:-)

[aristiguieta]

By the way, there are not so many clothing items that qualify, so if we get more answers a workflow like this would benefit from making sure we do ten (at least a second) pass on the list of options.

Not to elegant but simple: Get 10 threads of the eligible items list, and "union" them into a large list before applying the "running totals".

[jssandom]

$360 to play with.. decided to spread it out with one of each item.. 37 items purchased in total