Closed TrialsOnTrails closed 3 months ago
the ideal situation is using the same number of polynomials to approximate each data point, e.g., the test file has 214 ID, then we use 214 power to a x, 213... 1,0
nothing too significant between rounding or not rounding
It seems that the testing results in kaggle are in integer numbers but not floating point. An appropriate manner of rounding would help to reduce MSE.
This needs another branch for testing