Timeline for What is the benefit of breaking up a continuous predictor variable?
Current License: CC BY-SA 3.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 14, 2014 at 14:35 | comment | added | Scortchi♦ | In both of your first two examples, discretization is trying to bluff its way into the party by latching on to a bona fide guest. Don't be fooled. (1) If you want to model not having an open revolving credit line as a distinct class just use a dummy variable to indicate that condition & assign any constant value for average revolving credit balance. (2) If you want to treat certain extreme predictor values identically, as "big" or "small", truncate them; no need to muck about with the rest of the values. The 3rd case is uncontested - feel free to add examples. | |
| Oct 14, 2014 at 14:15 | history | edited | Scortchi♦ | CC BY-SA 3.0 |
fixed typos
|
| Oct 14, 2014 at 14:02 | comment | added | Scortchi♦ | Dichotomization is putting into two bins - do you mean discretization? | |
| Oct 6, 2014 at 3:50 | review | Late answers | |||
| Oct 6, 2014 at 3:50 | |||||
| Oct 6, 2014 at 3:32 | review | First posts | |||
| Oct 6, 2014 at 4:42 | |||||
| Oct 6, 2014 at 3:31 | history | answered | cjthompson | CC BY-SA 3.0 |