A procedure whereby a forecast of a disaggregate component is based on the forecast made of an aggregate variable (e.g., a forecast of menthol toothpaste based on a forecast for all toothpaste). Although this approach loses information about trends in the components (e.g., menthol flavor is becoming popular), reliability is usually better. (See also <a href="search.pl?p=bottom-up">bottom-up</a>.) <o:p></o:p></P>