Single variables to interactions

Stop scoring one factor at a time and ask which factors only matter in combination. Some causes do nothing alone and everything together, so a description that tracks each separately will keep missing the pattern that actually predicts the outcome.

Stop scoring one factor at a time and ask which factors only matter in combination. Some causes do nothing alone and everything together, so a description that tracks each separately will keep missing the pattern that actually predicts the outcome.

Example

Deadline weeks alone are fine. Your travel alone is fine. But deadline week plus you traveling is when every fight happens. The variable that predicts your six worst arguments is the pair, not either one, so you track the overlap on a shared calendar.