Page 128 - Textos de Matemática Vol. 47
P. 128
118
M. E. SILVA
●
● Residual Arrival
Departure
●
● ●●
● ●
● ●●●●
●●●●● ●● ● ●● ●●● ● ●
●●●● ●●●● ●● ●●●●●●● ●
●
●●● ●●●● ●● ● ● ●●● ● ●
● ●● ●●●● ●●●●●
● ●●●●
●● ● ●●● ● ●●●●● ●● ●●● ● ●●●●●●● ●●● ●
●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ●
●●● ● ●●● ●●●●●●●●●
●●●●●● ●
●●●● ●●● ●●● ●● ● ● ●● ●
●●●
●● ●
●●●●●● ●●●●● ● ●●● ●●●●●●●●● ●●●●● ● ●
●●●●●●
0 50 100 150 200 t
Figure 5. IP data set: residuals of a PoINAR(1) with an outlier at time t = 224.
5. Final remarks
Time series of counts arise in a wide variety of fields. The need to analyse such data adequately led to a multiplicity of approaches and a diversification of models that explicitly account for the discreteness of the data. One approach is based on the generalized linear models theory for dependent data [29]. Another point of view into the problem is given by parameter driven models which postulate that the observed process is driven by an unobserved process [13]. Yet another approach to the problem of modelling dependent count data is based on the use of renewal processes for generation of a correlated sequence
Residuals
−1 0 1 2 3
