| Nonlinear Dynamics, Psychology, and Life Sciences, Vol. 5, Iss. 2, Apr, 2001, pp. 175-191 @2001 Society for Chaos Theory in Psychology & Life Sciences Attractor Reconstruction with Principal Components Analysis: Application to Work Flows in Hierarchical Organizations Abstract: Explored several procedural questions in the use of the principal components technique for filtering time series data and reconstructing attractors in two or more dimensions. Data originated from an earlier study on work flows in experimental hierarchical organizations. The reconstructed time series and attractors were evaluated alongside a nonlinear regression modeling approach for assessing the dynamical properties of data. The results showed that, if the nonlinear function representing the time series is initially poorly fitted, principal components filtering will probably help to identify a nonlinear structure insofar as it has been fed enough data to capture the complexity of the time series of events. The relative merits of the eigenvalue 1.00 criterion and the scree test for determining the correct number of components were also addressed. Oblique rotation procedures are recommended. Keywords: principal components analysis, attractor reconstruction, factor extraction, Lyapunov exponents, scree test, organizational hierarchies, work performance |