test case issue problem 86

[moe18]

https://www.deep-ml.com/problems/86
Test cases are not good,

I passed all cases with:
if abs(training_accuracy - test_accuracy) > .2: return 1

[SAHAYAASHIKLIBU]

I think no. Currently, there is five test cases. And those cases check all possible scenarios like over-fitting, under-fitting and good fit. In addition the problem is relatively simple so it's not possible to add more complex test cases ig.