This competition is gonna improve your dataset collection skills!
This competition is gonna improve your dataset collection skills!
Unlike most of the kaggle's competitions, here you won't have any training set, instead, are you gonna gather it
yourself
! Excited?
Moreover, there is no testing set! You have an access to
all of the data points
we have in the competition! Excited even more?
You may ask: "Where is the catch?". Here you go! There is
no target variable
. 
We have chosen a crazy non-linear 10 dimentional function, that a participant needs to approxmnate. In order to collect the dataset, the participants are proivded with an ap, that returns a target for an given set of variables. Example of an API call:
http://139.59.68.8/get_value/PERSONAL_API_KEY/?d1=0.630868739146&d2=2.11659566222&d3=1.08099106563&d4=2.48976813389&d5=2.66392202604&d6=2.71439830305&d7=1.14912205052&d8=2.76907394478&d9=5.62600470797&d10=5.964746185
example of response
{"request": {"d1": 0.630868739146, "d2": 2.11659566222, "d3": 1.08099106563, "d4": 2.48976813389, "d5": 2.66392202604, "d6": 2.71439830305, "d7": 1.14912205052, "d8": 2.76907394478, "d9": 5.62600470797, "d10": 5.964746185}, "response": 3.412720422204076}
The API has a delay, and to gather all of the data points, you will need to spend 25days. But active learning techniques allow you to act smart! You can build your own training simpler more effectively with fewer samples, reach the top and get cool machine learning T-shirts!
If you aren't excited now, I don't know when you are!
YALLA!