 Heteroscedastic

Observations or errors are heteroscedastic when their variances are not all equal.

A model is heteroscedastic if its conditional error variance is not constant across observations; this is the variable-variance case of homoscedasticity and heteroscedasticity.