@nadirn wrote:
I was following a 2 year old article on predicting bike share users from data using Machine Learning
In this method the author split up the hours category in 8 bins/categories. I was thinking about splitting it up into simply 24 bins instead. Other than the extra computation cost that this would add when running my algorithm, would this be bad in any way? I feel like it would only make the algorithm better. I am usually wary of adding more features when the number of data points are limited, but with 17,000+ data points, that is not a concern I have here.
In this method the author split up the hours category in 8 bins/categories. I was thinking about splitting it up into simply 24 bins instead. Other than the extra computation cost that this would add when running my algorithm, would this be bad in any way? I feel like it would only make the algorithm better. I am usually wary of adding more features when the number of data points are limited, but with 17,000+ data points, that is not a concern I have here.
Posts: 1
Participants: 1