Weekly Challenges
Solve the challenge, share your solution and summit the ranks of our Community!Also available in | Français | Português | Español | 日本語
IDEAS WANTED
Want to get involved? We're always looking for ideas and content for Weekly Challenges.
SUBMIT YOUR IDEA
Good old random forest, have to say the error is a bit large but got the job done.
Also had a go trying this in Alteryx Cloud Platform > Machine Learning and went for the random forest. Pretty smooth (see attached PPT with model metrics and top 5 in the xslx, seems to disagree though with the Designer Forest Tool)
Here's my solution. I'm slightly in disbelief at how easy this question is relative to predictive analytics in general.
Couple points that the answers seem to be skimming over:
- The problem asks about 'collections', not deliveries or other. (hint: filter by collections)
- The problem asks about which are most likely to 'unsuccessful' so those are the ones to be cancelled. (hint: sort order of most unsuccessful, not most successful)
- Nobody seems to be 'oversampling' (in truth under-sampling the successful, and the reality is the problem doesn't actually direct you to adjust sampling, though I did that here) used a 23.3% based on 'original data'
- Even with these caveats, I couldn't match the provided solution. (though provided solution appears to target the 'most' successful', not the most 'unsuccessful', which are the routes we're trying to identify to be canceled)
I followed a similar approach in solving this challenge as @RolandSchubert , @ahsanaali and most others. Thank you, guys, for setting the model, as Predictive tools is not my strongest suit by all means :)
