DATA ANALYSIS
| Collection | International Encyclopedia of Systems and Cybernetics |
|---|---|
| Year | 2004 |
| Vol. (num.) | 2(1) |
| ID | ◀ 818 ▶ |
| Object type | Methodology or model |
T. BAUMGARTNER et al. state: “A major task of data analysis is to assist in the choice between alternative models. The question arises then, under what conditions would the analysis of empirical data suggest a multilevel system” (1976, p.36).
Besides, different possible uses of the data may imply a choice between different models to organize them. A very general problem arises from the frequent necessity to use short or very short series of data to integrate a would be long-term model, in order to obtain some forecasting power. As shown by the Italian demographers C. GINI (1952, p.81-92) and C. CIPOLLA (1973, p.63-70), imprudent extrapolation can lead to quite absurd forecasts: A seemingly exponential growth at short term may very well be a logistic growth (or asymptotic, or other) when, later on, some control factors become more effective and visible.
We now moreover know that some processes are chaotic, which adds a new type of difficulty to select a really fitting model for data integration.