### Table 1. The 6 Candidate Fitting Models

"... In PAGE 4: ...data sets, thus paves the way for applying learning curves to data mining. 2 Candidate Models, Algorithms and Data Sets The six candidate models used in this study are briefly described in Table1 . All of them are non-linear and respectively belong to three different major function families: Power Law family, Exponential family and Sigmoidal family.... ..."

### Table 3.1: The 6 Candidate Fitting Models Model Name Notion Formula Remarks

2002

### Table 3. Values of the Bayesian Information Criterion (BIC) for each subject and model.

2006

"... In PAGE 10: ... By fitting all four model variants to individual subjects, we can quantitatively assess the differences in the BIC goodness of fit statistic for the candidate models. Table3... ..."

### Table 2. Summary of statistics from the candidate regression models

2002

"... In PAGE 6: ... 2: (i) x2, x3, x1x2, x1x4; (ii) x2, x1x2, x1x4, x3x5; and (iii) x2, x3, x1x2, x2x4. A simple RA was completed for each of the above se- lected subsets, and a summary of the relevant fitting statis- tics is presented in Table2 . Notice that in addition to the three criteria for model selection discussed in Section 2.... In PAGE 6: ... More importantly, a paired t-test was performed to compare the means of the fitted values and their cor- responding observed values, and an F-test was conducted to compare the variances of the fitted values with those of their observed values in order to check the goodness of fit of each candidate model. It turns out that models I-1, I-2 and I-3 in Table2 have a marginal P value for the F-test based on a 5% significance level or 95% confidence level, and models III-1, III-2, and III-3 are among the best qualified candidates. Models II-1, II-2, and II-3 also have a satisfactory goodness of fit based on the two kinds of hypothesis testing.... In PAGE 6: ... For example, MAD would pick model II-3, MAPE model II-1, MRSE model II-3, and MSEP models III-1, III-2, and III-3. Interestingly, none of them picked those models disqual- ified by the hypothesis tests during model fitting shown in Table2 .Actually, the models picked by the three traditional measures, MAD, MAPE, and MRSE, are not significantly different, because the MAPE value of model II-3 is only slightly greater than that of model II-1.... ..."

### Table 3. Summary of prediction error statistics from the candidate regression models.

2002

"... In PAGE 6: ... If the other fitting statistics are considered, models III-1 and III-2 have almost the same quality of fit, since their residual squared error, adjusted R2, PRESS value, and prediction R2 Prediction are all among the best. Finally, the prediction error statistics from CV were com- puted for the above model candidates and are provided in Table3 .Nounits are given in Table 3 since these prediction error statistics are based on the logarithmic transformation.... In PAGE 9: ... In addition, the P values of the paired t-test and the F-test are also provided in Table 5. Notice that the prediction error statistics in Table 5 were obtained when the 24 models from Table3 were applied for prediction with the respective independent validation data set. Among the free parameters in Table 5, the first two parameters are the used training and testing tolerances, respectively.... ..."

### Table 1: Permutation Model Fit for Advice-Seeking Behavior

2005

"... In PAGE 17: ...) One dyadic exchange was made per candidate move, and A was taken to consist of all permutations on VY . Table1 shows parameter estimates and fit information for the permuta- tion model on the untransformed data. For comparison, both the MPLE and MCMC-MLE values are shown; while both are fairly similar, the differences are large enough to be substantively meaningful.... ..."

Cited by 1

### Table VII. Model Fitting Results for the CBCL Aggression as Reported by the Teacher

2003

Cited by 2

### Table 1. Model parameters for the two disk shadow candidates.

2005

"... In PAGE 6: ...2. Model of CK 3 The best-fitting parameters of the CK 3 model are given in Table1 and the model temperature and density structure are shown in Fig. 4.... In PAGE 10: ... However, the central compact near-infrared source is a clear signature of a projected disk shadow. The best-fitting parameters for Ced 110 IRS 4 are given in Table1 and the temperature and density structures are shown in Fig. 9.... ..."

### Table 1: Candidate models

2004

"... In PAGE 4: ... The number of defects is de- termined by the model parameterizations and the current time interval. Table1 lists the models and research papers in which they are discussed. 3.... ..."

Cited by 9

### Table 1. Candidate models

2004

"... In PAGE 4: ... The number of defect occurrences within a time interval is modeled as a non-homogenous Poisson process with a stationary defect rate a0 (t). Table1 lists the models. Lyu [16] provides details about the models, including researchers who have developed and applied the models in practice.... ..."

Cited by 9