1

Assume I want to predict if I'm fit in the morning. One feature is the last time I was online. Now this feature is tricky: If I take the hour, then a classifier might have a difficult time with it because 23 is numerically closer to 20 than to 0, but actually the time 23 o'clock is closer to 0 o'clock.

Is there a transformation to make this more linear? Probably into multiple features? (Well, hopefully not 60 features if I do the same for minutes)

Martin Thoma
  • 19,540
  • 36
  • 98
  • 170

1 Answers1

2

The question was already posted, you can find the answer there :

What is a good way to transform Cyclic Ordinal attributes?

The idea is to transform your time feature into two feature : it's like if you represent the hour as the angle of the hand on the clock, and use the sin/cos of the angle as your features

Nicolas Dol
  • 66
  • 4