### Table 7. Estimation Results of the Neural Network

"... In PAGE 25: ...5. The estimation results of each model are shown in Table7 . In this verification, the data were not adopted as estimation objects, except for those data that were evaluated as the highest and lowest within a set of visual objects.... ..."

### Table 3: Results of the best neural network and multiple regression models

2005

"... In PAGE 26: ... The views elaboration of system is supported by the use cases description technique [9][7]. This technique consists in describing the use case as an action sequence that will make an agent to achieve his goal ( Table3 ) [4]. An action is an effect produced by an agent according the given way on the system or by the system itself.... In PAGE 27: ...Journal of information and organizational sciences, Volume 29, Number 2 (2005) 19 Table3 : Use cases decomposition in actions Agents Use case Decomposition in actions Loan a1: To identify an adherent a2: Count adherent (To check right loan for member) a3: To seek for an exemplary a4: To treat an exemplary (to validate the output of an exemplary) a5: To treat adherent (to indicate the loan by adherent) Reservation a1: To identify an adherent a2: Count adherent (To check right loan for member) a3: To seek for an exemplary a6: To reserve an exemplary Restitution a1: To identify adherent a3: To seek for an exemplary a4: To treat an exemplary (to validate the input of exemplary ) a5: To treat adherent (to indicate the return of an exemplary) Loan if counts blocked a1: To identify adherent a2: Count adherent (To check right loan for member) Librarian Identification member a1: To identify adherent New adherent a2: Adherent account (to Add adherent) Adherents responsible Litigation a1: To identify adherent a2: Count adherent (Blocked count adherent) a7: To inform adherent Exemplary addition a3: To seek for an exemplary a8: To add copy Exemplaries responsible Exemplary withdrawal a3: To seek exemplary a9: To withdraw the damage exemplary Figure 4: Viewpoint diagrams of MEDIA LIBRARY system After, the Analysis Profile must enable the analyst to acquire scenarios as collaboration diagrams for each use case. The Collaboration diagrams concentrates on the structure of interaction between objects and their inter-relationships rather than focuses the temporal ... In PAGE 72: ...f the matching algorithms. Its function was in the evaluation of match results. This data set was set up especially to evaluate the field-matching effectiveness of the various algorithms irrespective of other constraints such as field weights. The results from running the three algorithms on this data set are presented in Table3 . Figures 3 and 4 depict the percentage precision and percentage recall achieved for varying thresholds respectively.... In PAGE 73: ...47%. Table3 : Experimental Results on Test data 2 Positional Algorithm Recursive Algorithm with word-base Recursive Algorithm with character-base Threshold %Recall %Precision %Recall %Precision %Recall %Precision 0.... ..."

### Table 1: Parameter estimates from linear logistic regression model and from average RGWs of neural network model.

"... In PAGE 8: ...hat the neural networks (a) outperformed the linear logistic regression model by between 2.5-7.5%; and (b) that the parameters used for the ANN models were adequate, so the interpretation results would be valid. Table1 presents estimators from a linear logistic regression model and averages of RGW over all patients with their sample standard deviation, and the estimation error for each covariate. The neural network results for all the continuous variables are very similar to the coe cients of the logistic regression models for these variables.... In PAGE 8: ... The RGWs for the binary variables were much smaller on average, suggesting that the ANN model was di erent from the linear logistic model, certainly with respect to the e ects of these covariates. Table1 about here.... ..."

### Table 4. Percent MAE for Neural Networks and Regressions on Test Set.

"... In PAGE 10: ... Therefore 60 regressions were created - 4 for the standard hold back approach and 56 for the leave-k-out. The estimation errors on the test data for these regression models and their neural network counterparts are shown in Table4 . For the leave-k- out models, the standard deviation of the estimations over the set of 15 (11 for the Drain Line Test) are also shown.... ..."

### Table 2: Comparative Predictive Capabilities of Regression and Neural Network Models

in Enhancing our Understanding of the Complexities of Education: "Knowledge Extraction from Data" using

"... In PAGE 17: ... The predicted outcome was compared with the actual for the non-sample schools and the MAPE values calculated across all ten random extractions. Table2 displays the results o f the prediction accuracy comparisons. As with the forecasting/time series study all neural networks significantly outperformed the linear regression models in this cross-sectional prediction exercise.... ..."

### Table 1: Comparative Predictive Capabilities of Regression and Neural Network Models

"... In PAGE 9: ...0 Results amp; Discussion 3.1 Predictive Accuracy of the Various Models Table1 displays the comparative prediction accuracy of all models tested. In the case of the high schools, five schools were randomly selected for prediction for each of ten trials, while with the larger elementary data set, ten schools were selected for each trial.... ..."

### Table 4 Prediction accuracies (LogR: logistic regression model, SVM: support vector machines, NN: neural networks) 10-fold

in Credit Rating Analysis With Support Vector Machines and Neural Networks: A Market Comparative Study

"... In PAGE 10: ... When performing the cross- validation procedures for the neural networks, 10% of the data was used as a validation set. Table4 summa- rizes the prediction accuracies of the four models using both cross-validation procedures. For comparison pur- poses, the prediction accuracies of a regression model that achieved relatively good performance in the liter- ature, the logistic regression model, are also reported in Table 4.... In PAGE 10: ... Table 4 summa- rizes the prediction accuracies of the four models using both cross-validation procedures. For comparison pur- poses, the prediction accuracies of a regression model that achieved relatively good performance in the liter- ature, the logistic regression model, are also reported in Table4 . The following observations are summarized: support vector machines achieved the best performance Table 4 Prediction accuracies (LogR: logistic regression model, SVM: support vector machines, NN: neural networks) 10-fold... ..."

### Table 2: Comparison between original object and object generated by neural networks.

2004

"... In PAGE 8: ... Three tables are drawn to show the differences between original z-values and estimated z-values using three criterions which are sum, average and standard deviation . Table2 -4 shows the calculated values as comparison. The result proves that 3D reconstruction using neural network is more accurate and compact than using the 3rd order polynomial.... ..."

Cited by 1

### Table 4: Neural Network Model Results

"... In PAGE 9: ...ean Abs. % Err. (MAPE) 19.52% Table 3: Regression Model Results As shown in Table4 , this specification provides a further improvement in model accuracy, with day- ahead MAPE values dropping to 17% and MAD values dropping to $4.5 per MWh.... ..."