Analysis of Update ID3 and the Proposed Method Based on WCY-IMD for 2008. 52

Chapter 6 Comparison of Decision Tree and the Proposed Method

6.3 Analysis of Update ID3 and the Proposed Method Based on WCY-IMD for 2008. 52

According to Table 6.1, and 6.2, update ID3 performs better than ID3 for all classes. To discriminate update ID3 and the proposed method, we take a data set from WCY-IMD for 2008 and induce rules for nation classes. The raw data is presented in Table B.4 and discretized data and classes in Table D.4 (nations are partitioned by k-mean of SPSS into 4

Table 6.6: Induced rules of Example 1 by the proposed method

Rule d

d

AR CR

R

(1): a

=1 and a

=2 0 0 0 1 0 0 0 1 0 1 1

R

(2): a

=2 0 0 0 0 1 0 0 0 0 1 1

R

(3): a

=1 and a

≥2 0 0 0 1 0 0 1 1 1 0.67 1

classes). The induced results are presented in Table 6.7.

Table 6.7: Comparison of update ID3 and the proposed method

Clas

s Rules by update ID3 AR CR Rules by the proposed method AR CR

G

a

= 4 0.78 0.5 (3 ≤ a

≤ 4) ∧ (3≤ a

≤ 4) ∧

(3 ≤ a

≤ 4) ∧ (3≤ a

≤ 4) 0.8 1

G

a

= 3 0.84 0.36 (3 ≤ a

≤ 4) ∧ (3≤ a

≤ 4) 0.80 0.84

G

a

= 2 0.82 0.45 (2 ≤ a

≤ 3) ∧ (2≤ a

≤ 3) 0.72 0.78

G

a

= 1 0.9 0.7 a

= 1 0.9 0.7

The comparison of induced rules for nation classes is presented below.

(i) The good rules generated by update ID3 can also be found by the proposed method.

For instance, induced rule for G1.

(ii) Some rules generated by the proposed method cannot be found by update ID3 such as G4, G3, and G2. The reason for this is that ID3 suffers from too many branches to give a quality rule. Users can take a look at Figure 6.5 and find 12 sub branches that are too many to cover enough nations to support rule quality.

Figure 6.5: The decision tree of nation classes for 2008

In summary, the advantages of the proposed method over the ID3 are listed below:

z The decision tree technique is a widely used method of inducing rules. However, its

S/a

a2=4

a

2=3

a

2=2

a

2=1

S

/a

S4/a10 S3/a13

S

4 3 2 1 4 3 2 1 4 3 2 1

induced rules may not be optimal, and may not cover all set of rules. That means the accuracy rate and the coverage rate for the rules found by the decision tree may not be the best.

z The greedy algorithm of the decision tree recursively partitions a tree with the most entropy reduction but it cannot guarantee the selected attribute is the best classifier.

z As described in the literature of Zanakis 2005 and Quinlan 1986, the decision tree method may expand many branches due to multiplication of the number of attribute values and those of classes. The more branches, the fewer nations are covered in a branch.

Chapter 7 Concluding Remarks

This research constructs an optimization model for inducing a nation’s dynamic rules based on the MCI-WCY data set. The rules, expressed in ’IF...THEN’ forms, are generated with a high coverage rate and accuracy rate. Based on the simple and consistent rules during 2001~2005, policy makers can use them to imply strategic initiatives for global competition or validate their decisions of economic policy.

The integer programming is applied to design models of rule extraction of competitiveness classes. Induced rules are composed of conjunctive and disjunctive terms of attribute values that give high coverage and accuracy rates. A visual displaying sphere is provided to show the dissimilarity of nations on the surface and also gives direction of competitiveness. Users can get a visual understanding of knowledge in WCY over years.

Advantages from two approaches of analyzing national competitiveness, ranks and classes, are combined into our proposed method. Stakeholders can catch competitiveness differentiation of classes and get explanation from induced rules, which directly point how to sustain, improve, or prevent fall in competitiveness. The features of the proposed model are listed below, compared with the other studies of inducing the rules of a nation’s competitiveness:

(i) Instead of using the WCY data set with a huge number of attributes, a high quality data set is used in the proposed model. This data set contains 14 most reliable and consistent attributes for 46 nations and three time periods, which help us to induce more reasonable rules.

(ii) Instead of using regression, neural network and decision tree techniques, the proposed model utilizes optimization techniques to induce rules. The rules have higher accuracy rates and coverage rates, and can be expressed in conjunction and disjunction terms.

(iii) Various types of national competitiveness have been found in this study, labeled as upward, downward, and sustaining, which are quite helpful to understand the critical factors affecting national competitiveness.

Some useful suggestions for both nations and investors are as follows:

(i) For nations wanting to move to a highly competitive level, the nations need to have a leading technology infrastructure or high export capacity.

(ii) For nations wanting to move from medium to competitive level, the nations need to have a medium level of GDP and have a high degree of technology infrastructure or stock market value.

(iii) For nations wanting to prevent falling in competitiveness, the nations need to lower unemployment rates and prevent GDP growth rates from declining.

(iv) GDP plays a fundamental role in forming nations’ classes.

(v) Computer, which indicates the level of technology infrastructure, is an essential attribute of competitiveness. A nation can enhance significantly its competitiveness by enhancing its computer network.

(vi) Export is another sufficient condition for a nation to be highly competitive.

(vii) A nation should prevent its inflation from being at the low level if it does not want to be the least competitive.

(viii) A nation of the least competitive or non-competitive class should enhance its Foreign Direct Investment and its export volume.

The limitations of this study and validation with data are discussed in the followings.

z A limitation of this study is the classification of nation groups. By applying the k-mean technique based on nations’ annual competitiveness scores, this study divides nations into 4 classes. Since k-mean is a heuristic method, the ranges of the 4 groups may not be the optimal.

z The attribute discretization (Appendix C) is determined approximately by equalizing the number of nations in each level. A more precise process of discretization may be studied in the future.

z The competitiveness scores before 2001 are not available, which restricts the time period of our study.

Appendix A: Classification of Nations

Here we use the algorithm of k-mean (reference) to classify 46 nations into four classes,

G

, G

2, and G1, based on the score values in TableB.1, Table B.2, and Table B.3. The steps are listed below (Han , 2001).

Step 1: Initialization

To assign initially each nation into a class for a specific year. Two criteria are used in the assignment. Firstly, the number of nations in each group is kept the same as possible.

Secondly, the gap of scores for the nearby groups is obvious. Take 2001 for instance, the initial classification is listed below.

2001

G

1 ={PHILIPPINES, INDIA, SOUTH AFRICA, ARGENTINA, TURKEY, RUSSIA, COLOMBIA, POLAND, VENEZUELA, INDONESIA}

2001

G

2 ={ JAPAN, HUNGARY, KOREA, MALAYSIA, GREECE, BRAZIL, ITALY, MAINLAND CHINA, PORTUGAL, CZECH REPUBLIC, MEXICO, THAILAND}

2001

G

3 ={ ICELAND, AUSTRIA, DENMARK, ISRAEL, BELGIUM, TAIWAN, UNITED KINGDOM, NORWAY, NEW ZEALAND, SPAIN, CHILE, FRANCE}

2001

G

4 ={USA, SINGAPORE, FINLAND, LUXEMBOURG, HONG KONG,

NETHERLANDS, IRELAND, SWEDEN, CANADA, SWITZERLAND, AUSTRALIA, GERMANY}

Step 2: Computing the average score,

C for a specific year y and class k, in each class.

_k^y

Following Step 1 the related

C values are as follows:

_k^y

2001

C

1 = 34.5, C2001

2001

C

2 = 48.8, C2001

2001

C

3 = 65.5, C2001

2001

C

4 = 81.4

Step 3: Denote

d

^y( k

i

, )as the similarity distance of nation i to the center of the class k, defined as

d

^y(

i

k

)=|

S

_i^y −

C

_k^y | where

S is the score of i

_i^y ^th nation in year y

Compute )

d

^y( k

i

, for i = 1, 2, 3, ..., 46 and k = 1, 2, 3, 4.

Step 4: If

d

^y(

i

k

)≤

d

^y(

i

k

^') then nation i belongs to class k.

Step 5: Recalculate

C and )

_i^y

d

^y( k

i

, until converges to a final solution. The final classification is listed in Table A.

For instance, ARGENTINA belongs to G1 (least competitive) in 2001, 2003, and 2005.

The group type is G₁₁₁. AUSTRIA belongs to G4 for 2001 and 2003, but downward to G₃ in 2005. The group type is G443.

Table A: Classified nations

2001 2003 2005

Nations

score classes score classes score classes

Competitiveness

Appendix B: Data Set of MCI-WCY

Table B.1: Raw data set for 2001

Nation

a

₁

a

₂

a

₃

a

₄

a

₅

a

₆

a

₇

a

₈

a

₉

a

₁₀

a

₁₁

a

₁₂

a

₁₃

a

₁₄ 46 VENEZUELA 795.7 63.3 4981 87 282.2 2.5 12778.0 16.2 10.2 120.0 19.6 3.2 503.0 2859.3

Table B.2: Raw data set for 2003

Nation

a

₁

a

₂

a

₃

a

₄

a

₅

a

₆

a

₇

a

₈

a

₉

a

₁₀

a

₁₁

a

₁₂

a

₁₃

a

₁₄

1 ARGENTINA 758.0 68 2875 111 4988.3 1.8 9496.0 25.9 17.8 83.2 -3.1 -11.2 241.2 1648.1 2 AUSTRALIA 3970.5 630 19965 1334 18411.2 13.8 42471.0 3.0 6.3 215.9 571.1 3.8 -873.6 11603.5 3 AUSTRIA 13566.2 476 25002 114 2976.9 20.4 53573.0 1.8 4.1 716.9 370.6 0.5 -201.7 14167.7 16 HONG-KONG 35021.6 434 24003 857 72925.1 5.2 50541.0 -3.0 7.3 3426.5 1634.0 2.3 2519.2 13098.0 17 HUNGARY 4163.7 153 6186 57 1031.7 2.1 16515.0 5.3 5.6 242.1 33.7 3.3 -257.9 4156.7 26 LUXEMBOURG 59891.3 460 45972 52 51739.1 18.3 76964.0 2.1 2.4 18760.9 1.0 3869.6 19130.4 27 MALAYSIA 4197.9 137 3814 809 4592.4 2.8 9900.0 1.8 3.5 21.0 10.3 4.2 275.9 1603.5 36 SINGAPORE 34931.0 596 20906 386 26965.5 7.8 43109.0 -0.4 4.4 2517.2 2195.4 2.2 4299.8 8597.7 37 SOUTH-AFRICA 721.3 84 2293 542 2981.4 5.3 7897.0 10.1 29.4 152.7 -78.7 3.0 6.7 1379.8 38 SPAIN 3997.7 224 16282 1458 10614.4 11.5 40229.0 3.5 11.4 488.3 628.0 2.0 -379.7 8589.9 39 SWEDEN 11327.1 687 26921 285 25701.7 19.5 56143.0 0.5 4.9 1446.4 769.1 1.6 1106.1 12817.7 40 SWITZERLAND 15157.3 641 36937 263 70053.8 23.7 64151.0 0.7 3.0 1159.9 1485.2 0.1 4279.6 21814.5 41 TAIWAN 6612.2 314 12500 584 12850.2 6.1 29777.0 -0.2 5.2 180.5 240.7 3.5 1130.0 7804.1 42 THAILAND 1252.9 43 1955 449 557.7 0.9 3759.0 0.7 2.2 58.7 2.5 5.2 117.2 1096.9 43 TURKEY 709.2 52 2631 310 654.9 3.5 9045.0 45.0 10.6 45.4 6.9 7.8 -24.8 1698.3 44 UNITED-KINGDOM 6400.7 526 25894 1923 36930.4 16.8 54775.0 1.6 5.1 1051.1 570.3 1.6 -217.0 16892.1 45 USA 3227.1 739 36552 6355 46574.9 20.3 77812.0 1.6 5.8 441.1 431.1 2.3 -1697.7 24622.0 46 VENEZUELA 1071.1 76 5137 63 233.3 2.5 14151.0 31.2 18.3 129.8 7.9 -9.6 287.5 3186.6

Table B.3: Raw data set for 2005

16 HONG-KONG 43786.7 503 23926 1029102968.3 5.5 50023.0 -0.4 6.8 4903.5 5727.7 8.1 2305.5 13876.1 17 HUNGARY 6251.0 191 9879 49 1617.1 2.1 25585.0 6.8 5.9 436.5 59.5 4.0 -873.0 6756.0 18 ICELAND 12266.7 726 41765 48 31000.0 19.1 78342.0 3.2 3.1 1466.7 8433.3 5.2 -3333.3 23666.7 19 INDIA 88.0 12 578 5644 251.9 0.6 1296.0 2.6 10.3 5.4 0.4 6.8 9.6 346.8 29 NETHERLANDS 26149.4 685 35629 183 30291.4 29.5 73074.0 1.2 4.7 973.3 2182.3 1.4 1475.5 17569.8 30 NEW-ZEALAND 5502.4 604 22369 157 8034.0 12.7 45358.0 2.7 3.9 623.8 118.9 4.6 -728.2 13155.3 36 SINGAPORE 48312.6 573 25191 475 33356.3 7.6 51680.0 1.7 4.0 3692.0 2452.9 8.4 6413.8 10367.8 37 SOUTH-AFRICA 1115.6 104 4574 426 5709.1 4.9 17782.0 1.4 27.8 12.4 34.3 3.7 -147.2 2874.8 38 SPAIN 5786.7 257 22968 3191 16463.4 16.4 57964.0 3.0 10.8 223.3 952.2 2.7 -1113.1 12931.3 39 SWEDEN 16722.7 741 38063 264 31768.0 27.7 79315.0 0.5 6.3 0.0 1319.3 3.3 3038.7 16254.1 40 SWITZERLAND 20215.1 712 48389 289 97540.3 30.2 85425.0 0.8 4.4 1918.0 3778.2 1.9 6733.9 28951.6 41 TAIWAN 8975.0 375 13459 669 16644.7 6.0 31204.0 1.6 4.4 83.4 311.4 5.7 834.4 8445.3 42 THAILAND 1741.7 57 2509 405 1823.6 0.9 4686.0 2.7 2.0 15.4 5.2 6.1 112.2 1396.5 43 TURKEY 1134.0 58 4190 284 949.1 3.5 13667.0 10.6 10.0 35.7 11.9 9.6 -215.1 2762.6 44 UNITED-KINGDOM 8144.7 595 35566 2311 40179.9 22.8 75779.0 3.0 5.0 916.1 1089.9 3.2 -504.7 19623.6 45 USA 3732.3 763 39468 5295 48112.4 22.0 84261.0 2.7 5.5 408.1 684.3 4.4 -2037.6 27759.3 46 VENEZUELA 919.5 90 4184 54 143.0 2.5 10667.0 19.2 16.6 42.9 -6.0 17.3 549.3 2061.7

Table B.4: Raw data set for 2008

Appendix C: Attribute Segments

Each attribute ai is divided into four levels with three cutting points P₁, P₂, and P₃. ai

belongs to 1^st, 2^nd, 3^rd, and 4^th level respectively if ai ≤ P1, P₁≤ ai ≤P2, P₂≤ ai ≤ P3, and P₃≤

a

i. The values of P₁, P₂, and P₃are determined by the approximate equal number of nations in each level. Take a1 for instance, by specifying P1=1645, P2=6537, and P3=13024, the number of nations at 1^st, 2^nd, 3^rd, and 4^th is 12, 12, 11, and 11. The cutting point of each attribute is listed in Table C.

Table C: Cutting point of indicator value

year

a

₂

a

₃

a

₄

a

₅

a

₆

a

₇

a

₈

a

₉

a

₁₀

a

₁₁

a

₁₂

a

₁₃

a

₁₄

p

1 2.645 179 6970 330 4.782 3 16310 3.3 8.3 0.316 0.157 3.9 0.064 4.486

p

2 6.537 408 20022 483 11.759 9 35405 4.5 11.8 0.645 .306 5.0 0.445 11.527

2001

p

3 13.024 523 26548 940 28.039 17 58320 8.0 15.3 1.467 1.053 7.2 1.206 14.344

p

1 2.457 126 4087 271 3.013 5 17764 4.2 5.0 0.228 0.469 2.2 -0.087 4.846

p

2 6.046 330 14103 397 10.295 10 38590 6.6 7.7 0.470 1.260 3.2 0.236 10.933

2003

p

3 10.833 593 25940 904 20.461 20 59416 11.4 10.9 0.956 2.021 5.1 1.041 15.466

p

1 3.437 237 6191 239 4.256 4 17743 2.5 5.1 0.269 0.295 3.6 -0.105 5.710

p

2 8.460 538 21628 352 12.482 9 45154 3.5 8.2 0.441 0.641 5.2 0.356 11.074

2005

p

3 15.157 688 37064 690 28.934 24 83531 5.2 10.2 0.785 1.161 6.8 1.509 20.013

p

1 82.4 232.4 9552 153 162 3.5 24643 2.03 4 3.1 2.2 3.1 -7.4 5811

p

2 216 558.8 27973 314 348.3 13 59121 2.76 6.1 10 12.3 5 5 13922

2008

p

3 435.8 717.5 43300 1165 1095.8 26 92955 4.9 8 32 43.9 6.5 28.9 23400

Appendix D: Discretized Codes of MCI-WCY

Table D.1: Discritized codes for 2001

Nation

a

₁

a

₂

a

₃

a

₄

a

₅

a

₆

a

₇

a

₈

a

₉

a

₁₀

a

₁₁

a

₁₂

a

₁₃

a

₁₄

Table D.2: Discretized codes for 2003

Table D.3: Discretized codes for 2005

Table D.4: Discretized codes for 2008

Appendix E: Data Set of Consolidated Competitiveness Factors

The data set of competitiveness factors is available for 2003 and 2005 but not for 2001. Its definition is described in Table 2.2. There are a few cells being empty which means no data available. Nations with empty cells are ignored on the displaying sphere.

Table E.1: Data Set of Competitiveness Factors

2003 2005 Nation

f1 f2 f3 f4 f1 f2 f3 f4

1 ARGENTINA 4.39 6.47 6.62 32.14 9.64 49.30 3.24 31.30

2 AUSTRALIA 89.91 57.16 85.79 82.48 73.39 53.03 78.62 65.05

3 AUSTRIA 64.91 59.19 68.67 73.47 58.89 50.79 68.94 60.87

4 BELGIUM 34.54 100.00 54.87 64.25 41.89 54.06 51.28 64.78

5 BRAZIL 36.45 34.00 49.67 33.35 19.96 45.64 48.87 27.13

6 CANADA 76.83 68.45 82.16 81.99 69.97 57.92 72.50 72.37

7 CHILE 66.38 44.59 63.50 26.60 68.22 52.13 76.96 33.60

8 MAINLAND-CHINA 56.57 73.86 36.98 33.93 58.57 70.94 28.59 36.95

9 COLOMBIA 52.35 32.29 43.76 38.86 41.88 36.46 39.18 30.29

10 CZECH-REPUBLIC 39.69 50.55 22.45 38.29 40.34 45.21 47.70 49.46

11 DENMARK 75.92 69.30 77.40 76.25 74.34 46.97 77.07 73.99

12 FINLAND 92.36 44.90 91.75 86.03 75.87 46.08 75.66 75.09

13 FRANCE 50.14 72.22 51.51 76.19 38.63 58.94 37.46 63.96

14 GERMANY 51.03 72.65 59.38 78.09 45.91 52.45 44.74 70.45

15 GREECE 17.59 31.22 34.63 27.00 31.12 40.18 31.08 41.13

16 HONG-KONG 85.14 69.46 89.18 48.19 83.29 70.11 98.60 62.47

17 HUNGARY 33.32 40.31 36.58 28.05 44.84 39.83 47.37 49.61

18 ICELAND 73.33 27.94 81.03 77.26 72.91 54.22 86.46 69.98

19 INDIA 41.48 52.68 43.53 27.24 42.83 56.85 53.34 25.38

20 INDONESIA 16.90 28.00 6.11 9.58 29.96 28.11 9.32 10.04

21 IRELAND 67.38 67.06 75.61 48.08 68.92 61.82 73.44 49.39

22 ISRAEL 21.68 48.33 60.47 47.86 43.93 55.74 63.87

23 ITALY 33.85 46.89 44.25 43.90 18.06 44.16 21.64 41.60

24 JAPAN 43.69 47.07 41.48 76.42 42.22 53.24 46.10 75.23

25 KOREA 43.24 39.32 42.07 50.00 47.57 42.48 49.21 59.88

26 LUXEMBOURG 80.20 80.43 74.01 55.88 66.5 77.22 60.84 58.87

27 MALAYSIA 78.02 63.51 69.79 60.51 51.22 59.53 51.11 43.69

28 MEXICO 43.72 42.23 26.11 21.32 33.70 41.11 17.98 15.32

29 NETHERLANDS 53.32 99.68 67.28 69.48 56.22 58.40 67.93 69.22 30 NEW-ZEALAND 69.07 47.27 61.44 52.53 72.61 54.61 63.44 53.36

31 NORWAY 59.14 62.32 54.95 70.10 64.17 50.26 60.50 71.88

32 PHILIPPINES 41.12 38.83 37.24 29.32 36.99 43.07 43.42 23.11

33 POLAND 15.00 23.37 17.09 30.13 21.22 35.49 11.47 30.06

34 PORTUGAL 46.77 43.92 6.47 20.14 42.20 42.40 25.12 42.16

35 RUSSIA 23.23 27.67 13.19 33.73 37.29 32.33 15.33 31.60

36 SINGAPORE 90.78 71.80 79.58 75.00 79.25 69.15 78.61 73.88 37 SOUTH-AFRICA 49.30 30.40 52.33 32.30 46.00 42.58 41.77 19.63

38 SPAIN 63.21 59.54 50.97 52.61 47.81 50.82 34.31 46.96

39 SWEDEN 62.84 68.21 67.36 84.56 57.95 49.21 67.61 72.46

40 SWITZERLAND 74.43 70.15 60.52 86.40 72.75 54.07 68.04 77.46

41 TAIWAN 63.32 52.71 74.77 64.17 60.19 54.19 77.19 63.89

42 THAILAND 70.86 66.85 53.49 34.29 64.51 59.69 50.58 31.45

43 TURKEY 16.96 11.96 45.19 34.86 31.70 37.10 50.91 27.64

44 UNITED-KINGDOM 61.53 71.39 58.17 60.54 51.03 56.50 51.01 57.72

45 USA 78.20 99.78 92.66 100.00 62.72 100.00 84.00 95.46

46 VENEZUELA 1.00 5.15 8.11 26.87 0.00 29.08 12.07 22.16

Table E.2: Coordinators of nations

2003 2005 Nations

x y z x y z 1 ARGENTINA -0.6572305 0.5637897 -0.5001892 -0.5261705 0.7468129 -0.4067125 2 AUSTRALIA 0.09157093 0.9610521 -0.2607557 0.2885919 0.8992801 -0.3286487 3 AUSTRIA 0.2462744 0.9452799 -0.2139973 0.1862301 0.8246496 0.3863444 4 BELGIUM 0.2585903 0.7992823 0.5424749 0.3443919 0.78848 0.3679042 5 BRAZIL -0.3454365 0.6381996 -0.6880224 -0.745025 0.60707 -0.2764125 6 CANADA 0.1367149 0.9758315 -0.1704756 0.2850063 0.9230719 -0.2582821 7 CHILE -0.3704571 0.9258821 0.07418797 0.08464952 0.7568797 0.5526248 8 MAINLAND CHINA -0.3906579 0.8974119 0.2050323 -0.3894844 0.7528272 0.628413 9 COLOMBIA -0.5311116 0.8403087 0.1086365 -0.7578207 0.5074652 -0.410106 10 CZECH-REPUBLIC -0.56091 0.7881293 0.2534406 -0.5639094 0.7994895 0.4131095 11 DENMARK 0.1736907 0.9697886 -0.1712944 0.3353658 0.8889409 -0.3119515 12 FINLAND 0.05729335 0.9429484 -0.3279722 0.344036 0.8854371 -0.3124744 13 FRANCE 0.3319612 0.8722968 0.3590266 -0.5771191 0.6684961 0.5562589 14 GERMANY 0.3044131 0.8762769 0.3734588 0.3517075 0.8028373 0.3206938 15 GREECE -0.4685574 0.6281605 -0.621183 -0.6937431 0.6205844 -0.3655073 16 HONG KONG 0.3415583 0.9346637 -0.09870023 0.06999692 0.9285804 -0.3644706 17 HUNGARY -0.5980036 0.7744642 0.2063899 -0.5752986 0.8093132 0.3808428 18 ICELAND 0.1899344 0.9016449 -0.3885373 0.2632078 0.9103639 -0.3193105 19 INDIA -0.2823032 0.719731 -0.6342651 -0.4130363 0.9013866 0.4235596 20 INDONESIA -0.5595774 0.7134025 -0.4218175 -0.6041083 0.5303078 -0.5948334 21 IRELAND 0.08414 0.9440161 0.3189891 0.2042338 0.8962 -0.3938453

22 ISRAEL 0.2757363 0.8296223 0.3131161

23 ITALY -0.6104489 0.782709 0.121321 -0.6258356 0.6896589 -0.3642805 24 JAPAN -0.6535853 0.7542793 -0.06236187 0.3751869 0.8018456 0.2961202 25 KOREA -0.593876 0.8013543 0.07171204 -0.6129484 0.7803701 0.3702867 26 LUXEMBOURG 0.301206 0.9513805 -0.06442118 0.1036893 0.9393901 -0.3267948 27 MALAYSIA 0.1230456 0.9562889 0.2652759 -0.4580393 0.8218026 0.4999124 28 MEXICO -0.3645716 0.7534457 -0.547181 -0.6763596 0.5226432 -0.51902 29 NETHERLANDS 0.2703884 0.9592568 0.08195455 0.3473483 0.9185222 -0.1888549 30 NEW-ZEALAND 0.1634185 0.9676746 0.1920947 0.1078082 0.8233171 0.4182776 31 NORWAY 0.277824 0.9115898 0.3030146 0.1856751 0.8533636 0.3027366 32 PHILIPPINES -0.3537547 0.7016286 -0.6185265 -0.7729034 0.5138644 -0.3722415 33 POLAND -0.5519935 0.6357766 -0.5395287 -0.5982569 0.6455617 -0.4746987 34 PORTUGAL -0.3958973 0.80583 -0.4403446 -0.6225334 0.7062649 0.4619567 35 RUSSIA -0.5235797 0.6771713 -0.5170138 -0.6520786 0.5734609 -0.4959194 36 SINGAPORE 0.1249795 0.9695915 -0.2104101 0.1797798 0.9489459 -0.2591929 37 SOUTH AFRICA -0.5004563 0.8540775 0.1417572 -0.7885693 0.4557103 -0.4129003 38 SPAIN -0.4844516 0.8742718 -0.03091253 -0.5411012 0.7619742 0.4926721 39 SWEDEN 0.235003 0.9623271 -0.1367481 0.3919433 0.8937901 -0.2179902 40 SWITZERLAND 0.226548 0.9614085 -0.1561079 0.3176676 0.9130052 -0.255947 41 TAIWAN 0.2232414 0.951489 0.2117355 0.3527142 0.9035925 -0.2431321 42 THAILAND -0.3830562 0.9237239 -0.00144001 -0.3656246 0.8588725 0.5255369 43 TURKEY -0.4913716 0.5350697 -0.6872076 -0.7805634 0.5361609 -0.321329 44 UNITED KINGDOM 0.206657 0.9137979 0.3496659 0.2889694 0.7896665 0.4101328

45 USA 0 1 0 0 1 0

46 VENEZUELA -0.6679537 0.5680158 -0.4808283 -0.4843282 0.6940127 -0.5327049

References

1. Adams, F. G., Gangnes B. and Shachmurove Y., Why is China so competitive?

measuring and Explaining China's competitiveness, The World Economy, Vol. 29, No.

2 (2006), pp. 95-122

2. Au, K. F., Wong, W. K., and Zeng, X. H., Decision model for country site selection of overseas clothing plants, International Journal of Advanced Manufacturing Technology, Vol. 29, SpringerLink, (London, 2006), pp. 408–417.

3. Feldstein, M., Argentina’s fall – lessons from the latest financial crisis, Foreign Affairs, Vol. 81 Issue 2 (2002), p8-14.

4. Garelli, S., Competitiveness of nations: the fundamentals, World Competitiveness Yearbook, IMD, Lausanne Switzerland, 2003.

5. Han, J., Cluster Analysis. Data mining concept and techniques. Morgan Kaufmann Publishers, San Francisco, (2001), pp.349–351.

6. International Institute for Management Development (IMD). World Competitiveness Yearbook, 2001.

7. International Institute for Management Development (IMD). World Competitiveness Yearbook, 2003.

8. International Institute for Management Development (IMD). World Competitiveness Yearbook, 2005.

9. Kruskal, J. B., By optimizing goodness-of-fit to a non-metric hypothesis, Psychometrica Vol. 29 (1964), pp. 1-27

10. Kruskal, J. B., Nonmetric multidimensional scaling: A numerical method, Psychometrica Vol. 29 (1964), pp. 115-129.

11.

Li, H. L. and Chen, M. H., Induction of multiple criteria optimal classification rules

for biological and medical data, Vol. 38 (2007), Computers in Biology and Medicine, pp.42-52.

12. Mak, B. and Munakata T., Rule extraction from heuristics: A comparative study of rough sets with neural networks and ID3, European Journal of Operational Research, Vol. 136 (2002), pp.221-229.

13. Mascitelli, B. and Gerstman, J., Italy's readiness for the 1998 Euro in view of the country's declining competitiveness, European Review, Vol. 16, no. 1 (Feb 2008), pp.

5-22.

14. Oral, M. and Chabchoub, H., Theory and methodology on the methodology of the World Competitiveness Report, European Journal of Operational Research, Vol. 90 (1996), pp. 514–535.

15. Pawlak, Z., Rough set approach to knowledge-based decision support, European Journal of Operational Research, Vol. 99 (1997), pp. 48–57.

16. Pistorius, C. W., Has South Africa fallen into a competitiveness trap, South African Journal of Science, Vol. 97 (2001), pp. 9–15.

17. Porter, M. E., Building the microeconomic foundations of prosperity: finding from the business competitiveness indicators. The Global Competitiveness Report 2005-2006, 2005.

18. Quinlan, J.R., Induction of decision trees, Machine Learning, Vol. 1, (1986), p81 –106.

19. Quinlan, J.R., C4.5: Programs for machine learning. Morgan Kaufmann Publishers.

CA, 1993.

20. Sheng, L., China and the United States: asymmetrical strategic partners, Washington, Quarterly, Vol. 22 (1999), pp. 147–164.

21. Sikder, I. U. and Munakata T., Application of rough set and decision tree for

Applications, Vol. 36 (2009), pp.102-110.

22. Wang, D. and Jiang, L., An improved attribute selection measure for decision tree induction, Fourth International Conference on Fuzzy Systems and Knowledge Discovery (IEEE), 2007.

23. Ulengin, F., Ulengin, B., and Onsel, S., A power-based measurement approach to specify macroeconomic competitiveness of countries, Socio-Economic Planning Sciences, Vol. 36 (2002), pp. 203–226.

24. Zanakis, S. H. and Becerra-Fernandez, I., Competitiveness of nations: A knowledge discovery examination, European Journal of Operational Research, Vol. 166 (2005), pp. 185–211.

在文檔中國家競爭力的規則推導與呈現 (頁 61-84)