Is it possible to use numbers less than zero in machine learning and scikit learn in particular?

For example, I need for ease of reading to determine the value of the team's win (.41) and loss (-.23), or when converting a continuous variable to ordinal (2 and -1, respectively).

How can such an interpretation affect the performance of the algorithm?

  • one
    You can use it, it should not affect performance, but it can affect the quality of predictions (although it all depends on the task) - mymedia

0