community
cancel
Showing results for 
Search instead for 
Did you mean: 
Do you have the skills to make it to the top? Subscribe to our weekly challenges. Try your best to solve the problem, share your solution, and see how others tackled the same problem. We share our answer too.
Weekly Challenge
Do you have the skills to make it to the top? Subscribe to our weekly challenges. Try your best to solve the problem, share your solution, and see how others tackled the same problem. We share our answer too.
Unable to display your progress at this time. Please try again a little later, or contact an administrator if you continue to see this error.
Getting started with Designer? | Start your journey with our new Learning Path!
SOLVED

Challenge #127: US Grand Prix Lap 2 - Employee Retention

Alteryx Certified Partner

It is clearly described how the models should be entered, however, no employee is awarded a predicted probability to leave above ~.35 so none of the records join (using 2019.2 here, so perhaps some changes were made within underlying R packages such as C5.0, cart, partykit?)

 

Anyways, I did what was expected in the challenge and I wasted a lot of time to figure out why my results are different

 

Spoiler
127.png
Comet

Getting in some last minute practice ahead of my grand prix prelims next week!

 

Spoiler
Challenge 127.PNG
Alteryx Partner

Couldn't get the same answer, then saw the posts mentioning versions which I'm confident is the reason...

Spoiler
Challenge 127.PNG
Alteryx Partner

Challenge 127 is done!

 

Spoiler
challenge 127 JMS solution.PNG
Asteroid

As others using newer versions, I couldn't get the same results... Definitely spent too much time trying to fix that

 

Spoiler
image.png
Highlighted
Bolide

Had it built in under 10 minutes, but took a long time to figure out why I was getting no results >= .5.  Finally looked at other newer replies and found the need to revert Decision Tree to earlier version.

 

Spoiler
Workflow 127.JPG

 

Asteroid

image.png

I slightly reinterpreted the challenge. Because of tool version discrepancies I decided to use all the available predictive tools to see if I can get similar records with any of them. The answer is no. I created my own list of employees by unioning the result from all the predictive tools and keeping the ones which came out of more than one tool.

 

 

Asteroid
 
Meteor
Spoiler
2019-11-14_15-29-42.png

My predictive tools aren't outputting >= 0.5 probability. Either are other solutions I've downloaded... probably the difference in Alteryx versions since this challenge is 1+ years old. I'll still chalk this up as a win?!