this hata is way harder than 4 kyu it deservs a 2
i have tried countless ideas and the more i think about it the more i m further away from the solution
i have solved 2kyu katas but this stands on a different level
the only solution i was able to impliment and that it actually worked is by random
in other words i wrote a programm that test all the posible connection and see wich one have the most while beeing a valid one
the problem in it is that it can t solve striongs that have more than 20-30 connections as it needs to go throu billions of possibilities and it falls in overFlowError
if any one have a solution for this kata i m more than gratfull to hear

Fixed

Added, if we missed any imports please let us know

Fixed

Needs import blocking.

Hardly the biggest issue when there's no module banning.

It's ill posed task, as it doesn't have any description of input dataset. It's just a luck it's modelled by linear regression. What is this task supposed to train/practice?

Quite mathematical challenge to figure out, useful kata for data science and machine learning, well done.

You're not just creating a model with this one - you are also choosing a model, and then testing data against it. Which makes it a bit more complex.

See this & this

done in latest fork

AttributeError: module 'sys' has no attribute 'stdout'

Reraised as issue

Just tried out this solution of yours, which is roughly equivalent if I am not wrong. Had 3 time outs in a row, and then it passed the test (in more than 11 s.). I don't know if tests should be adjusted to that level of precision to match between languages (it's a bit hard because performance can change with updates). My opinion is that for a 5 kyu it could be reasonable to require more efficient solution. But, should be Python's tests updated too in this sense? That would invalidate passed solutions... I don't know what should be done, it'd be nice to have more opinions about this.