### Table 2: The performance of various regression techniques

"... In PAGE 3: ...Table 2: The performance of various regression techniques Table2 shows the results for the 3 classes of regression algorithms we tested. Our first observation is that even simple linear regression outperforms our 2 level model by 4%.... ..."

### Table 2 Parameter estimates using linear regression

2000

"... In PAGE 9: ...simple linear regression to determine the parameters of the model [Jai91]. Table2 contains the numerical values of the model parameters for the three base stations considered separately and when taken together. We note that the values for the path loss exponent (n) and the reference signal strength (Pdo) for all three base stations are similar despite their different physical locations and surroundings.... In PAGE 9: ...Table 2 Parameter estimates using linear regression The final column in Table2 shows the values for Pdo and n when the data from all the transmitter-receiver pairs (i.... ..."

Cited by 675

### Table 2 Parameter estimates using linear regression

2000

"... In PAGE 9: ...simple linear regression to determine the parameters of the model [Jai91]. Table2 contains the numerical values of the model parameters for the three base stations considered separately and when taken together. We note that the values for the path loss exponent (n) and the reference signal strength (Pdo) for all three base stations are similar despite their different physical locations and surroundings.... In PAGE 9: ...Table 2 Parameter estimates using linear regression The final column in Table2 shows the values for Pdo and n when the data from all the transmitter-receiver pairs (i.... ..."

Cited by 675

### Table 2 Parameter estimates using linear regression

2000

"... In PAGE 9: ...simple linear regression to determine the parameters of the model [Jai91]. Table2 contains the numerical values of the model parameters for the three base stations considered separately and when taken together. We note that the values for the path loss exponent (n) and the reference signal strength (Pdo) for all three base stations are similar despite their different physical locations and surroundings.... In PAGE 9: ...Table 2 Parameter estimates using linear regression The final column in Table2 shows the values for Pdo and n when the data from all the transmitter-receiver pairs (i.... ..."

Cited by 675

### Table 1. Analyses of body mass index, gender, and age in controls from Genair/EPIC and in the EPIC cohort EPIC*, c (b) Controls from

"... In PAGE 7: ... It is possible that cross- sectional results in controls will produce a quite biased representation of associations in the overall cohort at baseline. The issues of bias and precision are illustrated in Table1 . The association between body mass index and gender and age was analyzed in the controls from Genair/EPIC and the corresponding portion of the EPIC cohort in which Genair/ EPIC was nested.... ..."

### Table 2: Exponential Regression of Single Technique Effectiveness

"... In PAGE 7: ...8 1.0 Resources Effectiveness Functional Testing Structural Testing Code Reading Figure 2: Single Technique Effectiveness on Ntree To validate the Exponentially Diminishing Model, an ex- ponential regression was performed against this effectiveness data, the results of which are given in Table2 . This non- linear least squares regression was performed in the R sta- tistical environment [13].... ..."

### Table 1: Linear Regression with f(x)=

"... In PAGE 16: ...Table 1: Linear Regression with f(x)= P k i=1 x i Some typical results for the simple linear regression are given in Table1 . There were nine simulated data sets as indicated in the table, chosen to explore the range of lowto moderate R 2 values.... In PAGE 17: ... Considering that the actual data is probably never really linear, and the errors are usually not really normal, the K-optimal test is probably preferable as a method for detection of association in this range of data. Table1 also shows evidence on the relation between and K suggested in the intro- duction. The four samples lin0 through lin3 are generated exactly in the same way except is decreasing from 3 to 1.... ..."

### Table 3: ANOVA table associated with simple linear regression analysis of data sets 1 and 2.

2005

"... In PAGE 7: ... These values relate to the regression items from the analysis of variance tables which are pro- duced in the original regression analysis conducted on dependent variable (y) that is accountable by the inde- pendent variable (x). From Table3 , the Regression SS for data set 1 was 217.6061, and the Regression SS for data set 2 was 190.... In PAGE 7: ... = -5.5815x + 42.864 R2 = 0.9863 15 20 25 30 35 40 12345 log (copy number) Mean Ct Data set 1 Data set 2 Data set 1 Data set 2 Page 7 of 13 (page number not for citation purposes) data sets 1 and 2, as illustrated in Table3 . The Regression SS characterises the component of the variation in the summed across both data sets, and SS[x] is the Sum of i=... ..."

### Table 3 shows that C5.00 is signi cantly less accurate than C5.0 on eight datasets ( rst column, last row) and signi cantly more accurate on ve ( rst row, last col- umn). It is signi cantly less accurate than M50 on seventeen datasets and signi - cantly more accurate on three. These results show that the superior performance of M50 is not due to the change in input/output encoding. We complete our discussion of the rst question by comparing simple linear regres- sion (LR) to M50 and C5.0. Table 3 shows that LR performs signi cantly worse than M50 on seventeen datasets and signi cantly worse than C5.0 on eighteen. LR outperforms M50 on eleven datasets and C5.0 on fourteen. These results for linear regression are surprisingly good. However, on some of the datasets the application of linear regression leads to disastrous results and so one cannot recommend this as a general technique.

1998

"... In PAGE 7: ... Throughout, we speak of results being \signi cantly di erent quot; if the di erence is statistically signi cant at the 1% level according to a paired two-sided t-test, each pair of data points consisting of the estimates obtained in one ten-fold cross-validation run for the two learning schemes being compared. Table3 shows how the di erent methods compare with each other. Each entry indicates the number of datasets for which the method associated with its column... In PAGE 8: ...5.0 in fteen datasets, whereas C5.0 outperforms M50 in four. (These numbers also appear, in boldface, in Table3 .) Of the sixteen datasets having numeric and binary attributes, M50 is signi cantly more accurate on nine and signi cantly less accurate on none; on the remaining datasets it is signi cantly more accurate on six and signi cantly less accurate on four.... In PAGE 10: ...FRANK ET AL. Table3 . Results of paired t-tests (p=0.... In PAGE 10: ...00 8 17 14 8 1 { To answer the second of the above two questions, we begin by comparing the accuracy of classi ers based on M50 with ones based on smoothed regression trees (SRT) to assess the importance of the linear regression process at the leaves (which the former incorporates but the latter does not). Table3 shows that M50 produces signi cantly more accurate classi ers on twenty-three datasets and signi cantly less accurate ones on only two. Compared to C5.... ..."

Cited by 20