Page 62 - Contributed Paper Session (CPS) - Volume 5

CPS886 Marcelo Bourguignon
We compared the BP regression model with the GA and RBS regression
models.

Figure 1: Scatterplots of nitrogen against productivity (a) and phosphate
against productivity (b).
Table 3 presents the estimates, standard errors (SE), Akaike information
criterion (AIC) and Bayesian information criterion (BIC) for the BP, GA and RBS
models. We can note that the BP, GA and RBS regression models present a
similar fit according to the information criteria (AIC and BIC) used.
Non-constant variance in can be diagnosed by residual plots. Figure 2(b)
note that the residual plot shows a pattern that indicates an evidence of a non-
constant precision because the variability is higher for lower phosphate
concentrations. Thus, we will consider the following model for precision of the
BP regression model
log( ) = + , = 1, . . . ,30.
2
1

0

The ML estimates of its parameters, with estimated asymptotic standard
̂
̂
̂
errors (SE) in parenthesis, are: = 0.5207(0.2788), = 0.3506 (0.0330), =
1
0
2
0.3990 (0.0423), ̂ = 2.7027 (0.6650) ̂ = 0.0072 (0.0033).
0
1
Note that the coefficients are statistically significant at the usual nominal
levels. We also note that there is a positive relationship between the mean
response (the productivity of corn) and nitrogen, and that there is a positive
relationship between the mean response and the phosphate. Moreover, the
51 | I S I W S C 2 0 1 9