Categorical data for binary classification

Question

ML newbie here. I'm preparing my data for a binary classification to predict whether a person has an account or not. In total I have 8 variables: 2 numeric (age and household size) and 6 categorical. Is it advisable to change my numeric variables, age and household, to categorical — i.e. age brackets — or better to keep them as discrete numeric values?

Thanks for your help and advice :)

Someone will post an answer or link to an existing answer about why this is a poor approach (you might get to learn of someone named Frank Harrell), but what advantage(s) do you see to binning your numerical variables? — Dave, Sep 10 '22 at 00:48
probably depends on the model you are using.. it might help to prevent overfitting, but you might not know how fine those intervals should be to keep the information — Alberto, Sep 10 '22 at 01:05

score 0 · Answer 1 · edited Sep 10 '22 at 14:02

0

Yes, you can do it, but be careful with the meaning of your content. All data can change from fine-scale (numeric) to coarse-scale (binary) but it will lose some detail.

edited Sep 10 '22 at 14:02

utobi

11,726

answered Sep 10 '22 at 01:45

YouKoEn

1

Welcome to Cross Validated! As the closure link shows, this sort of approach has major drawbacks. – Dave Sep 10 '22 at 15:07
this is helpful, thank you! – adam g. Sep 10 '22 at 17:23

Categorical data for binary classification

1 Answers1