### Table 1. Maximum likelihood binary classification performance for site multijects.

"... In PAGE 4: ... and forest representing vegetation and greenery. The detection performance of the five site multijects over the test-set is given in Table1 . The results in Table 1 are based on a maximum likelihood classification strategy.... ..."

### Table 2: Analysis of maximum likelihood estimates in logistic regression. Standard Chi-

1992

"... In PAGE 5: ... Nine di erent models are attempted in the analysis. Table2 summarizes the results of an analysis using the SAS procedure CATMOD [13]. For the pixel features, the results indicate that the decisions of both PNC and PBC... In PAGE 6: ...A set of 12000 samples are used to test the performance of these models. Table 3 shows the performance of each of the six classi ers and their combinations by the regression method with parameters given in Table2 . The parameter is set to zero if it is determined to be insigni cant.... ..."

Cited by 8

### Table 2: Total Variance Explained Component Extraction Sums of Squared Loadings

"... In PAGE 5: ... 3.3 Factor Analysis The percentage of variance explained was calculated by factor analysis, applying the maximum likelihood method and Varimax rotation ( Table2 ). The flrst two factors ex- plain 28.... ..."

### Table 1.|SNP Estimation { Maximum Likelihood Surface.

"... In PAGE 10: ... Thus, for a given SNP speci cation, the smaller the R-squared of the regressions, the better the SNP model approximates the true density. In Table1 we present the maximum likelihood surface for three key models: (1) the basic ARCH(1) model, which is SNP(11100); (2) the ARCH model with many lags in the variance (given by SNP(1h100), which has 17 lags in the variance) to approx- imate the GARCH(1,1) speci cation; and (3) the preferred model from our selection procedure, which is SNP(1c121). The preferred model is a general nonlinear process with heterogeneous innovations.... In PAGE 10: ...ith 816 observations for each of three series implying a saturation ratio of 26.3.9 Table 1 indicates that the preferred model performs substantially better than the other two models according to all three model selection criteria. (Insert Table1 here.) The superior performance of the preferred model in matching the data is also re-... ..."

### Table 4. Manufacturer Discretion: Maximum Likelihood Estimates

"... In PAGE 19: ... On their own, the average price sold, the number of dealers in the network and the age of the network explain around 70% of the variation in the allocation of completion rights by these contracts. [ Note: Table4 here, with allocation of rights by MLE] The regressors are also economically significant. An increase in price of automobiles of one standard deviation (Pta- 1.... In PAGE 20: ...88) increases manufacturer discretion by 1 clause. The main observation that can be derived from using MLE methods to estimate the relation between decision rights and network characteristics as showed in Table4 is that the results of such a procedure are entirely consistent with those in Table 3. None of the 28 signs of the dependent variables is altered by the change in methods.... In PAGE 20: ... Our confidence in these results is increased by the analysis of the individual clause variation presented in Table 5. 13 The signs of the individual effects are overwhelmingly the ones that Table4 has led us to expect: of 60 possible signs (15 regressions times 4 independent variables), only 4 are different than in Tables 3 and 4 and all of those 4 are insignificantly different than 0. As in Tables 3 and 4, particularly robust appear the results on Car Price, Number of Dealers in the Network, and the Asia dummy.... ..."

### Table 2: Maximum likelihood results for the superalloy data

"... In PAGE 8: ...where L i ( )= i flog[ (z i )] ; log[ (x i )y i )]g +(1; i )log[1; (z i )]: The ML estimate ^ of is the set of parameter values that maximizes L( )orL( ). Table2 gives the ML estimates of all model parameters resulting from tting the fatigue- limit model to the data. Figure 1 shows curves of the ML estimates of the 5, 50 and 95 percentiles of fatigue life.... In PAGE 9: ...stimators can also be computed. See pp. 292-297 of [2]. However, as mentioned earlier likelihood con dence intervals perform better in the sense that coverage probabilities are closer to nominal con dence levels than those of normal approximation intervals. The con dence intervalsin Table2 indicate that the parameters [ ] 1 , [ ] 1 and are statistically signi cantly di erent from zero. These intervals indicate that there is a rela- tionship between mean fatigue life and the stress level.... ..."

### Table 3: Classi cation accuracy of the Maximum{Likelihood classi er for each class.

"... In PAGE 8: ...WSG88] functions i.gensig and i.maxlik have been used to generate the training signatures and to perform the Maximum{Likelihood classi cation. The resulting error matrix is shown in Table3 , gure 4 shows the corresponding classi cation map compared to the ground{truth. The large time lag (6 years) between the completion of the vegetation map and the Landsat data take results in a degradation of the ground{truth accuracy.... ..."

### Table 3: Classi cation accuracy of the Maximum{Likelihood classi er for each class.

"... In PAGE 7: ...WSG88] functions i.gensig and i.maxlik have been used to generate the training signatures and to perform the Maximum{Likelihood classi cation. The resulting error matrix is shown in Table3 , gure 4 shows the corresponding classi cation map compared to the ground{truth. The large time lag (6 years) between the completion of the vegetation map and the Landsat data take results in a degradation of the ground{truth accuracy.... ..."