### Table 11. Class A sites identified as outliers by participants in the EMDI I workshop No. Latitude Longitude BIOME2 Site name, country Comments 1 -20.27 148.12 Grassland Bowen Pertusa,

"... In PAGE 48: ... A total of 11 records were flagged resulting in a total of 151 records for 81 unique sites. Virtually all the sites identified at the EMDI 1 Workshop ( Table11 ) as outliers were included in this set. We plotted NPP observed against the modeled NPP ensemble by AET ensemble and by latitude with the plot symbols indicating the magnitude of the flags.... ..."

### Table 8. The tables show that with method 1, the three axes that were computed without using the outlier, P1, are similar to each other and di erent from the axes computed using P1. Thus, it is possible to identify P1 as an outlier and ignore axes corresponding to this dot when the average is computed. With method 3 it is not possible to detect an outlier when four dots are considered. Hence, if outliers are possible, method 1 can be useful in spite of the necessity of weighting.

"... In PAGE 14: ...88 0.95 Table8 : The vector s, representing the direction of the axis, computed with method 3 for di erent triplets of dots when p11 is an outlier.... ..."

### Table 2. Comparison of the outlier detection methods

"... In PAGE 10: ...Observe that the M estimator based robust regression approach does not identify outliers. Table2 provides values of the three measures used to assess the methods as well as the regression equation parameter values computed in the way explained above. The obtained results show that the multilayer per- cpetron is the best technique for categorizing the data, followed by the PCA based approach.... ..."

### Table 1 we identified and removed View 17 and View 19 as outliers for the RF class. After removing these outliers the distribution of the RF class conforms to a normal distribution as illustrated in Figure 4. This is also confirmed by applying a Lilliefors test.

2006

"... In PAGE 3: ... Our initial focus was to investigate whether a statistically significant difference of this fractal dimension exists among the two classes. Table1 illustrates the actual calculated regularization dimension for each subject and both classes. We investigate the possibility of overlap between the two classes due to the fact that no malignancy existed for any of the subjects, regardless of the presence of galactographic findings.... In PAGE 3: ... Table1 . The fractional regularization dimension for every subject in each of the two classes.... ..."

Cited by 1

### Table 7 shows the results obtained from the scenario II simulation, using Geometry 1 with a set of different angular-rate aiding options. From Table 7 it can be seen that if the multi-antenna GPS antenna is aided with a roll or pitch rate gyro or an INS, then the correct outlier can be identified and removed from the solution for scenario II (described previously). Whereas a GPS only or a GPS combined with a heading rate gyro is able to detect that an error has occurred, it is unable to identify the correct

in Quality Control Issues Relating to an Attitude Determination System using a Multi-Antenna GPS array

2002

"... In PAGE 14: ... This confirms the results obtained from separability and localizability (see tables 4 and 5). Table7 Results for detecting outliers with angular rate gyros included in the solution for Geometry 1 with an outlier of 0.14m introduced to observation D1 (scenario II).... ..."

Cited by 7

### Table 5. Results for Wisconsin breast cancer data according to outlier factor

2002

"... In PAGE 8: ... When one in every six malignant records was chosen, the resultant data set had 39 (8%) malignant records and 444 (92%) benign records. The results are shown in Table5 . Within the top 40 ranked cases (ranked according to the Outlier Factor), 30 of the malignant cases (the outliers), comprising 77% of all malignant cases, were identified.... ..."

Cited by 16

### Table i. BrainMap outliers. The entries are ordered according to novelty. The second column indicates the paper, experiment and location identifier of the BrainMap database. The third to fifth column are x, y and z with the reported coordinates from BrainMap (not the corrected Talairach 1988 coordinates).

### Table A.1: BrainMap outliers. The entries are ordered according to novelty. The second column indicates the paper, experiment and location identifier of the BrainMap database. The third to fifth column are x, y and z with the reported coordinates from BrainMap (not the corrected Talairach 1988 coordinates).

### Table i. BrainMap outliers. The entries are ordered according to novelty. The second column indicates the paper, experiment and location identifier of the BrainMap database. The third to fifth column are x, y and z with the reported coordinates from BrainMap (not the corrected Talairach 1988 coordinates).

2001

### Table 1. Outlier detection on NHL(03/04) data

"... In PAGE 9: ... The first test mines the outliers based on the three attributes: games played, goals scored and shooting percentage. LOF, Orca and our reference-based approach achieve identical results and the top three outliers are listed in Table1 (a). The out- lier status of the three identified players are obvious.... In PAGE 9: ... The second test is to mine outliers based on the three attributes: points scored, plus-minus statistic and penalty minutes. The top 3 outliers found by our reference-based approach are listed in Table1 (b). Sean Avery is on top be- cause his points and plus-minus figures are moderate but the number of penalty minutes is the highest among all the play- ers.... ..."