I mean, big ensembles or stackings are also not useful in production but for some reason they count as "okay" solutions

Quite unfair

Well, finding this feature exactly was a critical thinking, in my opinion

@ravinder Yes, I am so much surprised that so called big men (i mean experienced Data scientists) in this Data Science journey could still be debating something like this that's not debatable at all, still arguing on something that virtually every Junior Data Scientist must have known before now. Why using a unique feature like "user_id" as one of the features for modelling. I think we don't need to be told at all. In fact, no need to even press further. Seriously still in shock hearing all these.

You are sounding quite toxic, tbh

1) There are no rules about id feature

2) @zindi does not even contact me to discuss it

3) There were no additional checks of my solution without this feature

4) It's not a million-dollar competition

Competitions are not about "making good production solutions" (and if you think so, I have bad news for you). It's more about education and fun. If you can not prepare data properly as a competition host or if you can not provide complete rules as a platform - it's not a participant headache. To be honest, it seems like @zindi just can't say something like "okay, fine, we complete our rules and also provide prize to 4th place" and it is just easier to ban me as a participant.

I also spent some time competing here, working with data and models and it is just unfair to me. I mean, I not even in leaderboard now.

"education and fun" ?? My friend, Zindi had previously provided many good data scientists in Africa and elsewhere and, to be honest, many people look up to Zindi to learn. If you are teaching people to use 'user_id' as a feature, I don't think it is in the best interest of anyone learning. Many young people who may not even be aware that using such features in their real jobs may be detrimental may learn this and suffer in the future.

"Competitions are not about making good production solutions" - Tell this thing to the organizers who trust Zindi with their problem statements and ask for resolutions from them. You are not the first one to get disqualified and I can fairly say with people using such practices in the future, you won't be the last.

Lastly, nobody is toxic to you or anyone over here. Everyone is stating facts. If you go to this link (https://zindi.africa/competitions/expresso-churn-prediction) and read properly you may find this written-

"Zindi is committed to providing solutions of value to our clients and partners. To this end, we reserve the right to disqualify your submission on the grounds of usability or value. This includes but is not limited to the use of data leaks or any other practices that we deem to compromise the inherent value of your solution."

Since we are here, nobody in Zindi is questioning your credibility as a data scientist/problem solver/hacker/whatever, just that you used something that was not in the best interest of the users and organizers, that's why Zindi took the decision. In fact, many people appreciated the other approaches you used as well. What is wrong needs to be pointed out and what's right must be appreciated and that's all we did.

Just take a lesson from it and move ahead as things tend to happen and we grow learning from it.

Toxic!!!! Nahhh.... Not at all. Why? Though, I was surprised to be hearing this, I must say. Was only talking on what we should all see as norm in this Data Science before now. If I may ask you a question. Have you ever seen any such rule being explained to you before on wether to use feature like "user_id" on Kaggle or any other platforms like this?? I will like to know.

@MICADEE here are some of them (only from Kaggle and only what I can remember):

Kaggle Predicting Red Hat Business Value: "people_id", "group_id"

Mercedes-Benz Greener Manufacturing: "ID"

TalkingData Mobile User Demographics: "rowID"

ASHRAE - Great Energy Predictor III: "building_id"

I also know few local competiotions where ids were also used, but they are in Russian so I'll not provide them

@AkashPB

1) My solution without this feature is almost production-ready (after refactoring)

2) With such a loose rules you can say "he was using second column" or almost everything else. I also don't remember it, but I read rules twice when I found this feature and there were no restrictions about data they had provided. I'm very sorry that I did not take a screenshot, because it seems to me that the text was edited exactly after my submission.

1. I did not say anything about your code being production-ready or not. Stop assuming.

2. Your this statement -

"I'm very sorry that I did not take a screenshot, because it seems to me that the text was edited exactly after my submission."

Well, what you are saying doesn't even make sense and now you are saying that text is edited after submission ?? Rules were pretty clear and I have been on this platform for over a year and these are the same set of rules I am seeing for a year. You can check all recently concluded competitions and all of them have the same rules, as far as I know.

1) You saying "Tell this thing to the organizers who trust Zindi with their problem statements and ask for resolutions from them." and my point is about it

2) I'm talking exactly about "or violated the spirit of the competition or the platform in any other way" and "disqualify your submission on the grounds of usability or value"

I want to clarify: I don't remember

All of this seems suspicious to me since they disqualified me. Try to put yourself in my position. I'm very upset about all of this.

1. Why tell them after you submitted your final solution with user_id. If that is the norm, I can in the future submit an overfitted solution with data leaks and tell the organizers that - oh sorry that was an honest mistake, kindly remove these features and use it. So your statement makes no sense here as well.

2. It was there right from start .. just scroll down the page, you will find it. It is not our decision. It is Zindi's decision.

@bbb Thank you for pointing this out. Because it was speculated there that ID must not be used and now that gives us grace to use "user_id" in this competition . Hmm..... I will have to stop here right now to avoid time wasting. We're deliberating too much on this 🤔.

But for clarity, you can check this out:

https://zindi.africa/competitions/indabax-nigeria-2021/discussions/7919frica/competitions/indabax-nigeria-2021/discussions/7919

Good luck to you next time bro.

Peace 👍

Link seems not working

But thank you (even though next time I'll probably compete on another platform)

Peace ✌️

@Micadee.....brother do not waste your time and energy in clarifying what seems to be very trivial and childish thing to me. If we state do not use ID in feature usage for any competition as a rule, it is just like we are giving a football to a kid and asking him explicitely to try hit only in the goal, but any decent kid would have that common sense to try hitting only in the goal, not at random place. We have clarified our opinion once-thats it. Use the same time to make one fruitful submission. Good day!!

True. Thanks@ravinder. Also, I am tired of explaining things that's expected to be known over and all over again. Mistakes are inevitable even though we should avoid the costly ones as best as we can, but still we should learn to accept it, learn from it as well. No big deal. This is not the end of the world.

Peace 👍.

@bbb You're welcome.

Just highlight the link and use your browser to open it. Pretty simple. The link is perfectly working.

https://zindi.africa/competitions/indabax-nigeria-2021/discussions/7919frica/competitions/indabax-nigeria-2021/discussions/7919

Inside, you will see all the messages I have been trying to pass across to you. No ill-feeling at all, I don't win any prize here as well. Let's move on.

Peace ✌️