"The separation of an intellectual or substantial whole into constituents for individual study" (The American Heritage Dictionary)
One of the most severe and persistent semantic muddle is the careless use of the term "analysis", even by systemists.
When a system is "analyzed", most or all of the nonlineal interrelations among its parts are unwittingly severed. As jokingly stated by J.L. LEMOIGNE, cutting a chicken into slices like a sausage does not yield an intelligible representation of the chicken. (1990)… and it should be quite more difficult to do.
Worse still, many people are not even aware of this problem. And different observers may "slice the chicken" in different ways, none very faithful to the original.
Analysis is perfect on its own (studying the components and at most their linear relations), but, as it generally leads to cutting the more complex interrelations between interacting parts, it is never a satisfactory substitute for the systemic global view, which is also indispensable for the study of complex systems.
Ideally, analysis should always be conjugated with synthesis, i.e. the synthetic view of the whole. Moreover this should also be a back and forth movement: analysis is useful and makes sense only in relation to the synthetic view and synthesis without a good knowledge of the interacting parts does not make much sense because it offers merely empty generalities that do not offer tools for action