|InterJournal Complex Systems, 580
|Manuscript Number: |
Submission Date: 20609
|Hybrid strategies for forecasting complex systems|
Subject(s): CX.0, CX.6, CX.4, CX.1
Category: Brief Article
This article considers the prediction problem in two disparate complex systems. The first is weather and hydrologic forecasting, the second is business forecasting and optimization. Both problems require reconciliation of information from disparate sources, which could vary by resolution, accuracy, predictive ability, and other attributes. Studies indicate that these problems share certain similarities in formulation and solution strategies. Data analysis and domain knowledge help to decompose the problems into components or processes. Hybrid modeling strategies can be designed, which apply domain knowledge or process physics and data dictated tools like traditional statistics or data mining, where each fits best. Well designed hybrid strategies can model the overall problem or the components thereof, and improve forecast skills over traditional approaches.
|Submit referee report/comment|