Skip to main content

Table 10 This table summarizes the results from the k − 1 ’LOOCV’ cross-validation analysis

From: Estimating the size of urban populations using Landsat images: a case study of Bo, Sierra Leone, West Africa

Name

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

Pop

Area

d

\(\sqrt{(}d)\)

\({\widehat{d}}\)

\(\widehat{d^2}\)

\(Er(\sqrt{d})\)

\(\%{Er(\sqrt{d}}\)

\(Er(\widehat{d^2})\)

\(\%{RE(\widehat{d^2})}\)

Roma

139

0.04

3510.72

59.25

70.39

4954.9

11.14

18.8

1444.18

41.14

Bo Central

273

0.066

4137.9

64.33

60.44

3653.53

3.88

6.03

484.37

− 11.71

Toubu

454

0.016

28089.66

167.6

160.05

25617.12

7.55

4.5

2472.54

− 8.80

Salina

580

0.467

1242.8

35.25

33.94

1151.93

1.31

3.73

90.87

− 7.31

Dodo

597

0.049

12126.22

110.12

111.22

12370.87

1.11

1

244.65

2.02

Reservation

637

2.329

273.5

16.54

16.91

285.8

0.37

2.22

12.3

4.50

Kpetewoma

640

0.197

3250.74

57.02

52.63

2769.82

4.39

7.69

480.92

− 14.79

Lewabu

879

0.479

1836.16

42.85

46.54

2166.06

3.69

8.61

329.9

17.97

Tengbewabu

1068

0.68

1571.17

39.64

42.41

1798.53

2.77

6.99

227.36

14.47

New York

1088

1.513

719.3

26.82

23.01

529.41

3.81

14.21

189.89

− 26.40

Komende

1103

0.196

5622.16

74.98

76.52

5856.04

1.54

2.06

233.88

4.16

Kindia Town

1160

0.146

7972.46

89.29

94.36

8903.89

5.07

5.68

931.43

11.68

New Site

1248

0.686

1818.17

42.64

47.27

2234.43

4.63

10.86

416.26

22.89

Yemoh Town

1858

0.404

4602.47

67.84

75.01

5626.04

7.17

10.56

1023.57

22.24

Median (abs)

1078.00

0.35

4370.19

66.09

67.11

4504.31

3.85

5.57

424.25

11.15

Mean (abs)

1297.70

0.48

6180.05

69.14

69.00

6098.19

3.89

7.41

568.21

14.58

  1. For each section, col (1) is the section population, Col (2) is the area in \(km^2\), and column (3) is the population density \(d = persons/area\). Col (4) is the transform \(\sqrt{d}\), col (5) is the estimate \(\widehat{\sqrt{d}}\), and (6) is the back-transformed estimate of \({\hat{d}}\). Cols (7) and (8) are the error (Er) and RE) of the estimated variable \(\widehat{\sqrt{d}}\); and cols (9) and (10) are the corresponding error functions for the back-transformed estimate of d