From the discussions in Section 5.4.3, we know that Eq. A2.5, the estimation that 
N 
minimizes (1/N) S60: —yu, es is the estimate that maximizes the likelihood function L, 
i=] 
or the logarithmic likelihood function log L. Accordingly, Eq. A2.8 can be written as 
AIC = — 2log(max likelihood) + 2(no. of parameters). (A2.9) 
Eq. A2.9 supplements the characteristics of AIC, this author believes. As another 
example of polynomial fitting, Fig. A2.3 shows the plot of the observed data for a 
maneuverability test of a ship. 
By the same procedure, from the behavior of AIC in Fig. A2.4, the order was 
determined as M = 5, and the fifth order polynomial was estimated as the best fit to these 
data. 
9 
a 
id 
> 
STARBOARD 
0.2 
YAWING RATE (DEG/SEC) 
° 
w 
HELM ANGLE (DEG) 
10 20 30 40 
STARBOARD 
ORDER 5 
ORDER 8 
Fig. A2.3. Fitting of polynomials to the data of maneuvering test of ship. 
(From Yamanouchi, et al.4°) 
264 
