Solutions to “Just KNIME It!” Challenge 11 - Season 4

Just KNIME It! JKI Season 4
21

Thanks for the thoughtful response!

I think we had a little “scope drift” in our thinking :blush:
I wasn’t aiming for a full brute-force optimization (that quattuordecillion figure made me smile :smile:), but rather something much lighter:

  • Sort by custom score or danceability
  • Go track by track, summing durations
  • If a track exceeds the remaining seconds → skip it and check the next one
  • Continue until you can’t fit more without exceeding 7200s

So basically: a greedy strategy with fallback. I agree it’s far from “perfect optimization”, but computationally feasible (300–400 rows max) and elegant enough for the challenge.

The Recursive Loop seemed the best choice because:

  • it tracks the growing playlist step by step,
  • handles table resizing dynamically,
  • and allows flexibility if we want to filter on danceability thresholds (e.g. ≥ 0.75).

So no Mars-level hardware required :smile:
Let me know if I misunderstood something too — always open to refining ideas!

22

Ahh I think I understand, so you would score it (not just using the danceability score) and sorting by your score, check where the 7200s mark is, aim for not reaching that, then start row by row checking if the next row in the rest table is fitting into the available space in the suggested table. :smiley:

I think it would be very elegant. And because of the dynamic table size changes (if one fits you want to check the other rows, so the table can grow) the recursive loop is the ideal choice for that.

I think your thinking is real great, now just the implementation awaits :smiley: (I have worked a lot with recursive loops so if you have any question I’m here to help :slight_smile: )

1 Like
23

Thanks — totally on point! :dart:
That’s exactly the logic I had in mind — a scored sort, stop at the cumulative limit, and then try to sneak in extra tracks one by one if they still fit. Recursive loop magic at its finest :jigsaw:

Now all that’s left is to convince my 4-year-old to give me a brief “research sabbatical” this evening :smile:
(Spending time with him is absolutely the best part of my day — but this tag-team KNIME brainstorming is a close second :wink:)

Seriously, I really appreciated this back-and-forth — the “tag team” approach made this challenge even more fun and insightful.

24

Ohh totally understand, I have two at home, a 4 and a 1 year old, and it’s kindergarten break (thank god) so everyone is at home :smiley:

Hang in with the implementation! :slight_smile:

1 Like
25

:sun_with_face: We just posted our solution to last week’s challenge on dance-y playlists! :dancer:

:robot: This challenge involved greedy optimization, moving aggregation, and ETL techniques. It was great to see some of you coming up with more than one playlist as an answer, and also some solutions including insightful graphics. :bar_chart:

:mantelpiece_clock: :alarm_clock: We hope to see you tomorrow for a new challenge – this time also including the processing of temporal data, but in the context of work progress tickets!