. 2013 Nov 18;8(11):e80962. doi: 10.1371/journal.pone.0080962

Table 2. Optimisation of the autocorrelation function, showing the auto-regressive (AR, φ _n) and moving-average (MA, θ _n) correlation parameters for models of increasing order.

AR order (p)	MA order (q)	φ ₁	φ ₂	θ ₁	AIC
0	0				1969.509
1	0	0.17668			1772.993
1	1	−0.07139		0.29979	1774.255
2	0	0.28949	−0.14868		1773.324

The AIC for each model was used to select the model which best described the error structure (shown in bold). As increasing complexity failed to produce models with a lower AIC, models with more than 2 auto-regressive and 1 moving average parameters were not run [39], [44].

Table 2. Optimisation of the autocorrelation function, showing the auto-regressive (AR, φ n) and moving-average (MA, θ n) correlation parameters for models of increasing order.

Table 2. Optimisation of the autocorrelation function, showing the auto-regressive (AR, φ _n) and moving-average (MA, θ _n) correlation parameters for models of increasing order.