Challenge #94: Have we reached Peak Pumpkin?

[JORGE4900]

Hello all,

Here is my attempt at this exercise; looking at the interactive lessons, I got an idea on how to import those workbooks much more dynamically. Other than that, I am pretty much following the approach everyone else is. 

https://community.alteryx.com/t5/Interactive-Lessons/Connecting-to-Multiple-Sheets-at-Once/ta-p/8287...

[JORGE4900]

Hello @Kenda

I really enjoyed looking at your macro.

It was a very dynamic approach to running the model.

Awesome!

[paul_houghton]

Thanks :)

[estherb47]

The only slight difference I took in the approach was to check if any weeks were missing as part of the workflow.

[LandonG]

Solution attached.

[PhilipMannering]

Fairly new to the predictive stuff. Here's my attempt. Tried it using the standard time series tools and the Factory tools from the gallery with my best guess at an explanation. Solution attached.

Spoiler

[Natasha]

Learned a lot while doing this challenge. didn't create a macro as I was excited to play with new Time Series Tools. My Produce numbers are also slightly off which I don't understand why, as at the first glance the solution approach is quite similar.

Spoiler

[ggruccio]

Had to use a few block until done tools to slow this one a bit.  Fun time series exercise!

[samN]

Had some trouble with downloading the complete predictive tools package. Am getting most of the tools but the ones i needed for this challenge were missing. Was able to copy them from the solution and then configure them how i needed . 

[SeanAdams]

So – after looking at other people’s solutions I realized that I did this in a very brute-force kinda way.

I blame this guy who was giving me advice (now I know the origin of that ancient saying “never take Alteryx advice from a lizard in a rainforest”)

Spoiler

• Data Input: I think most of us used the dynamic input to bring the data in with a few different approaches
  • @Kenda & @estherb47 & @patrick_digan used a directory tool into dynamic (few different approaches in here)
  • @JORGE4900; @LandonG; @NicoleJohnson; @PhilipMannering& I did the input tool to get sheets, then dynamic
  • @LordNeilLorddid the textInput for a list of years, then a dynamic
  • @ChristineBdid a directory tool, into a macro
• Check for missing weeks
  • @estherb47used a multi-row tool to check
  • Some others used the TimeSeries filler tool
• Prediction: Then in order to do the prediction there were a few variants too
  • @Kenda; @NicoleJohnson& @LandonG did a macro
  • Some golks just did ARIMA for one product
  • And the slick solution seems to be the TSFactory tools (which I never knew existed) - @patrick_digan; @ChristineB; @paul_houghton; @PhilipMannering; @LordNeilLord
  • I did it the brute-force way by having one Arima per product <very sheepish>
    
    

Solution Notes:
- My solution matches the forecast provided by @ChristineB in the solution for all products except Produce.   Strange – most of the values are within 1 of the target so probably a rounding error, but some are quite far off. 
- Also – the predicated sales for a few products are negative (both in the provided solution and in mine) – that should be interesting to see!
Based on bad advice from a lizard in a forest