Forecasting, Structural Time Series and the Kalman Filter
Citations
303 citations
278 citations
Cites background from "Forecasting, Structural Time Series..."
...A tem’s elements, combined with the influence of a large set of external factors. The value of the tourism destination comprises a number of elements: the tourism operators, the support struc- chaos and complexity framework in understanding the development of a destination and the role of tures, public and private organizations and associations. All of these elements have some kind of small tourism business networks has also been discussed by Tinsley and Lynch (2001). relationship among themselves and the possible nonlinearities in these relationships are well known The main objective of this article is to give a and have been described several times (Farrell & brief overview of the complexity framework and Twining-Ward, 2004; Faulkner & Russell, 1997). to explore the implications and contributions that Moreover, we can include in the system also ele- the study of complex systems can give to the unments not traditionally thought as belonging derstanding of the tourism destination model. Constrictly to the tourism sector, but whose impor- tinuing the line of research presented above, this tance and role in this framework is undoubtedly work aims at complementing and reinforcing it by very high. providing some quantitative evidence in support An important, although rather scarce, strand of of this approach. This, it is hoped, will allow the literature has pointed out the necessity to change reader to gain a deeper appreciation of this point attitude when studying tourism and tourism sys- of view. tems. In a pioneering work, Faulkner and Valerio The remainder of the article is organized as fol(1995) start from the realization of the deficiencies lows....
[...]
...For ships is considered to be detrimental for the devel- example, Agostinho and Teixeira de Castro (2003) opment of the system, because evolution and analyze a Brazilian experience and provide tangigrowth can only be possible in regions of the ble data showing that an adaptive, self-organizing, phase space at the boundary between order and management system produce better performance...
[...]
...A time se- tions of the initial separation vector. Thus, there exists a whole spectrum of Lyapunov exponents. ries can be used to derive such a plot. Before doing that, we must recreate the phase space by using Their number is equal to the number of dimensions of the phase space. The largest LCE deterone of the techniques devised for this purpose. The most commonly used is the time-lagged (de- mines the general behavior of the system. If it is negative, the system follows a stable trajectory; if lay-coordinate) technique (Kantz & Schreiber, 1997; Schreiber, 1999). A delay coordinate recon- it is null, the system is in a steady state; if it is positive the system exhibits unstable and chaotic struction can be obtained by plotting the time series versus a time-delayed version of it. For a two- behavior (Sprott, 2003). In most cases, the calculation of Lyapunov exponents cannot be carried out dimensional reconstruction, it is possible to plot the delay vector yn = (tn, tn−V), where V is the lag or analytically and numerical techniques must be used. In cases like ours, when only a one-dimensampling delay: the difference between the adjacent components of the delay vector measured in sional time series is given, the highest LCE can be number of samples. The theoretical basis for this estimated with the method proposed by Wolf, procedure is due to Takens (1980). His fundamen- Swift, Swinney, and Vastano (1985) and Rotal theorem states that a dynamical system can be senstein, Collins, and De Luca (1993)....
[...]
...Harvey, A. C. (1989)....
[...]
...) transmission are greatly improved with re(1998) and Barabási and Albert (1999) have prospect to a random ER network, in some cases it vided evidence that, in many cases, real-world netis shown that there are no critical thresholds at works are quite different from ER graphs....
[...]
256 citations
Cites methods from "Forecasting, Structural Time Series..."
...(7.5) If mi signifies the last time point with observations, the optimal predictor of mT under this model is easily obtained by application of the recursive Kalman filter equations (Harvey 1989)....
[...]
...See Harvey (1989) for details....
[...]
...If m i signifies the last time point with observations, the optimal predictor of m T under this model is easily obtained by application of the recursive Kalman filter equations (Harvey 1989)....
[...]
254 citations
199 citations
References
5,260 citations
4,571 citations
1,497 citations
26 citations