Warhammer related ML question.
I’m a beginner CS student, just trying to pick up some intro machine learning practice. I’m trying to train a linear regression model (using SKlearn and Pandas on google colab).
For one of my input variables, it’s functionally on a scale of 2-6, or Null (2 being the best, 6 being the worst, and null being worse than the worst). Is it better to set the null inputs to zero, to set them to seven, or is there a way to leave them null?
For those who care, I’m messing around with training a MLM to ingest a Warhammer data sheet and predict its point cost. The thing in question here is the invulnerable save, where each attack that goes into a model is guaranteed to be blocked on a roll of equal to or higher than the stat. 2+ is the best, as it blocks 5/6 attacks, and 6+ is the worst. However, not all models get an invuln save, and having one is better than not having one.