### Table 1: Statistics for the 10 training sets. Set Period Mean Std. Skew. Kurt. Min. Max.

1999

"... In PAGE 13: ...ets of length 1100 with subsequent test sets of length 563. In Fig. 4 the training and test sets are indicated by dashed and dotted lines, respectively. The exact periods of time13 covered by the sets together with some basic statistics are available from Table1 and 2. In particular, the mean, the standard deviation, the skewness, and the kurtosis are reported for all training and test sets as well as the minimum and the maximum return.... In PAGE 23: ...hat the models with a non-gaussian conditional distribution, i.e. the models 3{6, are 19The common assumption underlying the paired t-test and the paired Wilcoxon test is that the di er- ences of the (paired) error measures are independent. Since the training sets are several years apart (see Table1 ), the error measures (and their di erences) on di erent sets can indeed be regarded independent. Additionally, for the paired t-test it is assumed that the di erences are normally distributed whereas for the paired Wilcoxon test it is only assumed that the distribution of the di erences is symmetric.... In PAGE 24: ... For this error measure, the non-linear models are signi cantly better than their corresponding linear models (with the exception of the paired t-test for the comparison of the models 3 and 4). The best models with respect to the hit rate HR are the RMDN(1) and the RMDN(2) models (see Table1 1). Note that a higher hit rate corresponds to a better performance... In PAGE 25: ...005 6: RMDN(2) 0.759 1 5 2 12 0 { Table1 0: In-sample statistics (NMAE): Mean values (second column), p-values for the paired t-tests (above the diagonal) and R-values for the paired Wilcoxon tests (below the diagonal). Model Mean 1 2 3 4 5 6 1: GARCH(1,1) 0.... In PAGE 25: ...006 6: RMDN(2) 0.692 20 22 16 15 2 { Table1 1: In-sample statistics (HR): Mean values (second column), p-values for the paired t-tests (above the diagonal) and R-values for the paired Wilcoxon tests (below the diag- onal). (on average).... In PAGE 26: ...773 6: RMDN(2) 0.700 20 19 19 21 16 { Table1 2: In-sample statistics (WHR): Mean values (second column), p-values for the paired t-tests (above the diagonal) and R-values for the paired Wilcoxon tests (below the diagonal). on average.... In PAGE 27: ....190. In fact our simulations (for data sets of length N = 1100) yield 3.898 (0.893) (see Table1 3) where the standard deviation is put into parentheses. For larger data sets (N = 104), the obtained estimates are better: 4.... In PAGE 28: ...076] 0.062 Table1 3: Additional statistics for set 1{5: for each set the rst line gives the values estimated directly from the training set whereas the second to seventh line give the mean values of the statistics calculated from 100 sets of the same length generated by the GARCH(1,1), RMDN(1), GARCH(1,1)-t, RMDN(1)-t, LRMDN(2), and RMDN(2) model, respectively.... In PAGE 31: ...064 4.275 Table1 5: Out-of-sample statistics (for set 1{5): The rst to sixth line gives the per- formance of the GARCH(1,1), RMDN(1), GARCH(1,1)-t, RMDN(1)-t, LRMDN(2), and RMDN(2) model, respectively.... In PAGE 32: ...656 6.043 Table1 6: Out-of-sample statistics (for set 6{10): The rst to sixth line gives the per- formance of the GARCH(1,1), RMDN(1), GARCH(1,1)-t, RMDN(1)-t, LRMDN(2), and RMDN(2) model, respectively.... In PAGE 33: ...561] 4.856 Table1 7: Estimated parameters and from Eq. (39) for the ten test sets for models 1{6: GARCH(1,1), RMDN(1), GARCH(1,1)-t, RMDN(1)-t, LRMDN(2), and RMDN(2).... In PAGE 34: ...258 6: RMDN(2) 1.128 13 4 10 23 17 { Table1 8: Out-sample statistics (loss function): Mean values (second column), p-values for the paired t-tests (above the diagonal) and R-values for the paired Wilcoxon tests (below the diagonal). Model Mean 1 2 3 4 5 6 1: GARCH(1,1) 0.... In PAGE 34: ...218 6: RMDN(2) 0.772 23 20 21 14 16 { Table1 9: Out-sample statistics (NMSE): Mean values (second column), p-values for the paired t-tests (above the diagonal) and R-values for the paired Wilcoxon tests (below the diagonal). p- and R-values for these measures are such that the di erences between the models are not signi cant.... ..."

### Table 3: Sample statistics of Example 5.1 Sample Size Mean Std. Dev. Ske. Excess Kurt. # of Limited Prices r t

"... In PAGE 15: ... To be comparable with the application in the last section, both rt and r t are rescaled by multiplying 103. Sample statistics of the simulated (observed and equilibrium) returns are reported in Table3 . In this example, the excess kurtosis of the unconditional observed return rt is also negative though the excess kurtosis of the equilibrium return r t is still positive.... ..."

### TABLE I COMPUTING TIME OF THE DEVICES (PENTIUM-IV-3000) USING DIFFERENT SIMULATED ROBOCUP RESCUE ARENAS. FOR COMPARISON: THE REAL KURT2 YIELDS ENCODER TICKS WITH A FREQUENCY OF 100 HZ, LASER SCANS WITH 75 HZ AND CAMERA IMAGES WITH ABOUT 10 FPS. Kurt3D device computing time (4 PC cluster) computing time (single computer)

2006

Cited by 3

### Tables 1 and 2 quantify the improvement on the previous bound for the Zipf distribution, where pi = 1=iHn. The Zipf distribution often governs the reference process, especially when keys are drawn from a text le (as in Lisp [2]). In Tables 1 and 2 we present the ratio between the stopping points achieved by the Chebyshev and Hoe ding bounds and m as computed in (16), respectively. As shown in Table 2, for any , the bound presented in Theorem 5 improves the value obtained by the Hoe ding bound by at least a factor of . In comparison with Chebyshev (Table 1), which also uses the variance for computing the bound, the improvement increases as becomes smaller. Acknowledgment: We would like to thank Micha Hofri, Kurt Mehlhorn and Ofer Zeitouni for helpful comments and suggestions.

### Table 1. Classification of junctions.

1996

"... In PAGE 3: ... When the system arrived at a junction, it performed a scan to acquire the actual data and have the network do the classification. Table1 shows the number of correct/uncertain classifications for all junctions. As these data show, the network is not sure to produce correct results.... In PAGE 3: ...he internal map (see Sec. 3) is used. In case of a discrepancy between how the network classified the junction and what is expected from the internal map, the classification procedure was retried: KURT would pull back into the pipe a short stretch, approach the junction, and classify again. Table1 also shows the results after retrying. Even though no incorrect classifications after retry were left in our tests, discrepancies may occur even then.... ..."

Cited by 34

### Table 7 - Monthly Returns for Different Countries During Their Phases of Economic

"... In PAGE 7: ... The data are available from the TSM Global database. Table7 presents the performance statistics for the individual phases for the different stock market indexes with all returns in US dollars. Table 8 provides the same results only in native currencies.... ..."

### Table 2. Coordinates of objects in the PSR B0656+14 eld.

1997

"... In PAGE 4: .... G. Kurt et al.: BVRI observations of PSR B0656+14 Table2 gives the coordinates of the pulsar optical can- didate, 6 bright stars (denoted by the numbers 1 through 6inFig.1),and4objectswithin 500 around the pulsar.... In PAGE 4: ... 3 and 4 that 5 objects are detected within lt; 500 around the expected pulsar position. The position of one of them (\o1,PSR quot; in Table2 and Fig. 6) coincides, within the budget of astrometry errors, with the radio position, so that this object is the most plausible candidate for the optical counterpart of PSR B0656+14.... In PAGE 5: ...+0.20;- 0.17) Table 5 gives estimates for magnitudes of these objects. These estimates and the distances of the objects relative to \o1,PSR quot; ( Table2 ) allow us to conclude that these... In PAGE 9: ... 6. Objects from nearest vicinity of the pulsar are indicated whose coordinates are given in Table2 , and estimates of stellar magnitudes are given in Table 5. Coordinates (in arcseconds) correspond to those in Figs.... ..."

### Table 2: unresolved hypotheses (excerpt)

1995

"... In PAGE 4: ... The procedure of proper name tagging descri- bed here is not able to recognize multi{word pro- per names because only two adjacent capitalized words (apposition + proper name) are examined. Table2 shows an excerpt of unresolved hypothe- ses in which some multi{word proper names con- sisting of rst name and surname (Albrecht M uller, Angelika Beer, Harry Ristock, Ruth Winkler, Josef Felder, Gabi Witt, Florian Gerster, Sepp Binder, Kurt Schumacher), of normal nouns ((das) Deutsche Rote Kreuz, Kleine Brogel, Ewige Lampe) and of some non-German proper names (Alan Stephenson, (Canadian) Air Group, Central Enterprise, Frecce Tricolori, Standardisation Agreement, Acrobatic Full Scale) are found. The non-German proper names are often put in quotation marks, so this could be an additional cri- terion for the hypotheses evaluation, but cases in which quotation marks are used to emphasize or to cite one or more words must be excluded (24).... In PAGE 5: ...27) ein Teil von ihnen geht [...] ins Gasthaus \Ewige Lampe quot; (some of them go to the inn \Ewige Lampe quot;) (28) ich habe in der Stra e \Am Mariahof quot; gewohnt (I have lived in the street \Am Mariahof quot;) Some of the remaining hypotheses in Table2 are noun pairs consisting of quantity terms and normal nouns (29)-(31) or constructions with month names (32). Quantity terms could be excluded by an ex- ception list and month names could be added to the EN{lexicon from the start.... ..."

Cited by 3

### Table 2.1 Performance of fleets according to data availability, average 2002-2004 *

2006

"... In PAGE 17: ... The output of this calculation is a percentage called the economic performance index which can be further categorized according to Table 2. Table2 . Table describing classification of break- even revenue in terms of economic performance.... In PAGE 35: ... A 22 m long selfspreading ground gear used as alternative to the 24/21 rockhopper ground gear illustrated on Figure 1. In Table2 is show some calculated forces (drag and spread) for three ground gear configurations, using a spreadsheet program developed for this purpose. The calculations are related to the ground gears shown on figures 1 and 4 recently tested onboard a Norwegian factory trawler (M/Tr Granit 4 ).... In PAGE 36: ... TRAWL DESIGNS FOR SHRIMP UTILIZING THE BEHAVIOUR CHARACTERISTICS OF THE TARGET SPECIES Behaviour observations of shrimp inside the trawl belly have demonstrated that the target shrimp are guided passively along the sloping belly of a trawl. Another important experience from commercial Innovations in trawl components that reduce the trawl drag soundrightraysrightJohn Willy Valdemarsen1, Kurt Hansen2 37 Table2 Calculated forces (drag and spread) for three ground gear configurations 22m plated gear 22 m rockhopper 21 8 m rockhopper gear (50X50 cm) + 2X 7m plated gear Towing speed Drag Spread Drag Spread Drag Spread (kn) (kg) (kg) (kg) (kg) (kg) (kg) 2 418 315 426 0 373 223 3 941 708 957 0 839 502 4 1673 1258 1702 0 1491 892 5 2614 1966 2659 0 2330 1394 Figure 6. The lifting kites as used during the ... In PAGE 71: ...000 200 0,04 0,20 Table 1: The composition of the Icelandic fishing fleet in the year 2004 Number of v essels A v er age po w e r of main engine (kW ) Figure 6: The composition of the Icelandic fishing fleet, number of vessels and average engine size versus type of vessel. Table2... In PAGE 72: ... The fleet is composed of four main types of ships; trawlers, big decked and small decked vessels and small boats. Table2 shows a summary of the energy usage of the fleet. The trawlers and big decked vessels are very energy consuming and stay at sea for consid- erably long time whereas the smaller boat use much less energy and comes much more frequently to shore.... ..."