第六章 結論與建議
B. 第二組資料執行結果
5-Fold Cross Validation Results ---
Accuracy Rate on Test Set | Rules Number | Conditions Number ---
83% +/- 7.68% | 2.8 +/- 0.2 | 2.6 +/- 0.4 Total elapsed time: 0 s.
B. 第二組資料執行結果
i. Weka 軟體執行結果內容
=== Run information ===
Scheme: weka.classifiers.bayes.NaiveBayes Relation: 060521test-1
Instances: 60 Attributes: 10 性別 年齡 教育程度 職業 婚姻狀況 宗教 血型 星座
是否喜歡喝酒 酒類
Test mode: 10-fold cross-validation
--- Cross Validation #1---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 啤酒: Prior probability = 0.17 Class 無: Prior probability = 0.69 Class 葡萄酒: Prior probability = 0.08 Class 威士忌: Prior probability = 0.03 Class 梅酒: Prior probability = 0.03
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
Correctly Classified Instances 54 90 % Incorrectly Classified Instances 6 10 % Kappa statistic 0.7662
Mean absolute error 0.084 Root mean squared error 0.1889 Relative absolute error 45.3916 % Root relative squared error 63.9773 % Total Number of Instances 60
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class 0.8 0.06 0.727 0.8 0.762 啤酒 0.977 0.063 0.977 0.977 0.977 無 0.75 0 1 0.75 0.857 葡萄酒 0 0.017 0 0 0 威士忌 0 0.017 0 0 0 梅酒
=== Confusion Matrix ===
a b c d e <-- classified as 8 1 0 0 1 | a = 啤酒 0 43 0 1 0 | b = 無
1 0 3 0 0 | c = 葡萄酒 1 0 0 0 0 | d = 威士忌 1 0 0 0 0 | e = 梅酒
--- Cross Validation #2---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 啤酒: Prior probability = 0.3 Class 無: Prior probability = 0.48 Class 梅酒: Prior probability = 0.03 Class 葡萄酒: Prior probability = 0.07 Class 威士忌: Prior probability = 0.06 Class 紹興酒: Prior probability = 0.03 Class 米酒: Prior probability = 0.03
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
Correctly Classified Instances 48 80 % Incorrectly Classified Instances 12 20 % Kappa statistic 0.668
Mean absolute error 0.0998 Root mean squared error 0.2195 Relative absolute error 53.4181 % Root relative squared error 72.9936 % Total Number of Instances 60
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class 0.947 0.146 0.75 0.947 0.837 啤酒 0.968 0.034 0.968 0.968 0.968 無 0 0.017 0 0 0 梅酒
0 0.018 0 0 0 葡萄酒 0 0.053 0 0 0 威士忌 0 0 0 0 0 紹興酒 0 0 0 0 0 米酒
=== Confusion Matrix ===
a b c d e f g <-- classified as 18 0 0 0 1 0 0 | a = 啤酒 1 30 0 0 0 0 0 | b = 無 0 0 0 0 1 0 0 | c = 梅酒 2 1 1 0 0 0 0 | d = 葡萄酒 2 0 0 1 0 0 0 | e = 威士忌 0 0 0 0 1 0 0 | f = 紹興酒 1 0 0 0 0 0 0 | g = 米酒
--- Cross Validation #3---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 無: Prior probability = 0.64 Class 葡萄酒: Prior probability = 0.14 Class 啤酒: Prior probability = 0.12 Class 高粱酒: Prior probability = 0.05 Class 梅酒: Prior probability = 0.03 Class 威士忌: Prior probability = 0.03
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
Correctly Classified Instances 54 90 % Incorrectly Classified Instances 6 10 % Kappa statistic 0.7987
Mean absolute error 0.0831 Root mean squared error 0.1888
Relative absolute error 46.542 % Root relative squared error 64.895 % Total Number of Instances 60
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class 1 0 1 1 1 無 0.75 0.058 0.667 0.75 0.706 葡萄酒 0.857 0.038 0.75 0.857 0.8 啤酒 0.5 0 1 0.5 0.667 高粱酒 0 0.017 0 0 0 梅酒 0 0 0 0 0 威士忌
=== Confusion Matrix ===
a b c d e f <-- classified as 41 0 0 0 0 0 | a = 無
0 6 2 0 0 0 | b = 葡萄酒 0 0 6 0 1 0 | c = 啤酒 0 1 0 1 0 0 | d = 高粱酒 0 1 0 0 0 0 | e = 梅酒 0 1 0 0 0 0 | f = 威士忌
--- Cross Validation #4---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 啤酒: Prior probability = 0.31 Class 無: Prior probability = 0.5 Class 藥酒: Prior probability = 0.03 Class 威士忌: Prior probability = 0.04 Class 梅酒: Prior probability = 0.03 Class 葡萄酒: Prior probability = 0.03 Class 米酒: Prior probability = 0.03 Class 高粱酒: Prior probability = 0.03
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class
=== Confusion Matrix ===
a b c d e f g h <-- classified as
--- Cross Validation #5---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 梅酒: Prior probability = 0.03 Class 啤酒: Prior probability = 0.23 Class 葡萄酒: Prior probability = 0.08 Class 無: Prior probability = 0.61 Class 高粱酒: Prior probability = 0.03 Class 藥酒: Prior probability = 0.03
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
Correctly Classified Instances 49 81.6667 % Incorrectly Classified Instances 11 18.3333 % Kappa statistic 0.6386
Mean absolute error 0.096 Root mean squared error 0.2271 Relative absolute error 52.2055 % Root relative squared error 76.7202 % Total Number of Instances 60
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class 0 0.034 0 0 0 梅酒 0.714 0.087 0.714 0.714 0.714 啤酒 0 0.071 0 0 0 葡萄酒 1 0.048 0.975 1 0.987 無 0 0 0 0 0 高粱酒 0 0 0 0 0 藥酒
=== Confusion Matrix ===
0 0 1 0 0 0 | a = 梅酒 0 10 3 1 0 0 | b = 啤酒 2 2 0 0 0 0 | c = 葡萄酒 0 0 0 39 0 0 | d = 無 0 1 0 0 0 0 | e = 高粱酒 0 1 0 0 0 0 | f = 藥酒
--- Cross Validation #6---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 啤酒: Prior probability = 0.28 Class 無: Prior probability = 0.45 Class 葡萄酒: Prior probability = 0.09 Class 白蘭地: Prior probability = 0.03 Class 紹興酒: Prior probability = 0.03 Class 米酒: Prior probability = 0.04 Class 藥酒: Prior probability = 0.03 Class 高粱酒: Prior probability = 0.03 Class 威士忌: Prior probability = 0.03
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
Correctly Classified Instances 42 70 % Incorrectly Classified Instances 18 30 % Kappa statistic 0.5244
Mean absolute error 0.0909 Root mean squared error 0.2131 Relative absolute error 59.5081 % Root relative squared error 78.6337 % Total Number of Instances 60
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class
=== Confusion Matrix ===
a b c d e f g h i <-- classified as
--- Cross Validation #7---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 無: Prior probability = 0.55 Class 啤酒: Prior probability = 0.26 Class 米酒: Prior probability = 0.05 Class 葡萄酒: Prior probability = 0.09 Class 高粱酒: Prior probability = 0.03 Class 藥酒: Prior probability = 0.03
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
Correctly Classified Instances 45 75 % Incorrectly Classified Instances 15 25 % Kappa statistic 0.5667
Mean absolute error 0.1112 Root mean squared error 0.243 Relative absolute error 54.8195 % Root relative squared error 77.5582 % Total Number of Instances 60
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class 0.914 0.08 0.941 0.914 0.928 無 0.75 0.159 0.632 0.75 0.686 啤酒 0 0.034 0 0 0 米酒 0.2 0.073 0.2 0.2 0.2 葡萄酒 0 0 0 0 0 高粱酒 0 0 0 0 0 藥酒
=== Confusion Matrix ===
a b c d e f <-- classified as 32 1 1 1 0 0 | a = 無
1 12 1 2 0 0 | b = 啤酒 0 2 0 0 0 0 | c = 米酒 1 3 0 1 0 0 | d = 葡萄酒 0 1 0 0 0 0 | e = 高粱酒 0 0 0 1 0 0 | f = 藥酒
--- Cross Validation #8---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 無: Prior probability = 0.45 Class 啤酒: Prior probability = 0.34 Class 威士忌: Prior probability = 0.03 Class 紹興酒: Prior probability = 0.04 Class 高粱酒: Prior probability = 0.04 Class 米酒: Prior probability = 0.03 Class 葡萄酒: Prior probability = 0.06
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class
=== Confusion Matrix ===
a b c d e f g <-- classified as 27 1 0 1 0 0 0 | a = 無
0 19 1 1 0 0 1 | b = 啤酒
0 1 0 0 0 0 0 | c = 威士忌 0 1 0 0 1 0 0 | d = 紹興酒 0 1 0 1 0 0 0 | e = 高粱酒 0 0 0 1 0 0 0 | f = 米酒 1 1 0 0 0 0 1 | g = 葡萄酒
--- Cross Validation #9---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 無: Prior probability = 0.56 Class 啤酒: Prior probability = 0.21 Class 威士忌: Prior probability = 0.04 Class 梅酒: Prior probability = 0.04 Class 紹興酒: Prior probability = 0.03 Class 葡萄酒: Prior probability = 0.04 Class 米酒: Prior probability = 0.04 Class 高粱酒: Prior probability = 0.03
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
Correctly Classified Instances 47 78.3333 % Incorrectly Classified Instances 13 21.6667 % Kappa statistic 0.6096
Mean absolute error 0.0862 Root mean squared error 0.2074 Relative absolute error 56.4671 % Root relative squared error 77.0234 % Total Number of Instances 60
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class
0.973 0.043 0.973 0.973 0.973 無 0.846 0.128 0.647 0.846 0.733 啤酒 0 0 0 0 0 威士忌 0 0.034 0 0 0 梅酒 0 0 0 0 0 紹興酒 0 0.017 0 0 0 葡萄酒 0 0.052 0 0 0 米酒 0 0 0 0 0 高粱酒
=== Confusion Matrix ===
a b c d e f g h <-- classified as 36 0 0 0 0 0 1 0 | a = 無
1 11 0 0 0 0 1 0 | b = 啤酒 0 2 0 0 0 0 0 0 | c = 威士忌 0 0 0 0 0 1 1 0 | d = 梅酒 0 1 0 0 0 0 0 0 | e = 紹興酒 0 1 0 1 0 0 0 0 | f = 葡萄酒 0 1 0 1 0 0 0 0 | g = 米酒 0 1 0 0 0 0 0 0 | h = 高粱酒
--- Cross Validation #10---
=== Classifier model (full training set) ===
Naive Bayes Classifier
Class 啤酒: Prior probability = 0.3 Class 葡萄酒: Prior probability = 0.1 Class 無: Prior probability = 0.6
Time taken to build model: 0 seconds
=== Stratified cross-validation ===
=== Summary ===
Correctly Classified Instances 57 95 % Incorrectly Classified Instances 3 5 %
Mean absolute error 0.0819 Root mean squared error 0.1751 Relative absolute error 23.0721 % Root relative squared error 41.7773 % Total Number of Instances 60
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure Class 1 0.071 0.857 1 0.923 啤酒 0.6 0 1 0.6 0.75 葡萄酒 0.973 0 1 0.973 0.986 無
=== Confusion Matrix ===
a b c <-- classified as 18 0 0 | a = 啤酒 2 3 0 | b = 葡萄酒 1 0 36 | c = 無
ii. Ant-Miner 軟體執行結果內容
=== Run Information ===
Relation: 060521 原始資料 Instances: 600
Attributes: 10 性別 年齡 教育程度 職業 婚姻狀況 宗教 血型 星座
是否喜歡喝酒 酒類
User-defined Parameters Folds: 10 Number of Ants: 10 Min. Cases per Rule: 10 Max. uncovered Cases: 10 Rules for Convergence: 10 Number of Iterations: 20
--- Cross Validation #1--- Cases in the training set: 540
Cases in the test set: 60 Rules: 9
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '女' AND 婚姻狀況 = '已婚' THEN '葡萄酒' IF 職業 = '以勞力為主之員工' THEN '啤酒'
IF 教育程度 = '專科或大學' THEN '啤酒'
IF 性別 = '男' AND 職業 = '負責人或部門主管' THEN '啤酒' IF 教育程度 = '國中或高中職' AND 血型 = 'O' THEN '啤酒' IF 性別 = '男' AND 血型 = 'B' THEN '啤酒'
IF 教育程度 = '國中或高中職' THEN '啤酒' Default rule: 高粱酒
Accuracy rate on the training set: 86.29629629629629 % Accuracy rate on the test set: 83.33333333333334 % Time taken: 5.266 s.
--- Cross Validation #2--- Cases in the training set: 540
Cases in the test set: 60
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '女' AND 婚姻狀況 = '已婚' THEN '葡萄酒' IF 宗教 = '無' THEN '啤酒'
IF 性別 = '男' AND 宗教 = '道教' THEN '啤酒' IF 婚姻狀況 = '未婚' AND 血型 = 'B' THEN '葡萄酒' IF 教育程度 = '專科或大學' THEN '啤酒'
IF 職業 = '以勞力為主之員工' AND 宗教 = '佛教' THEN '啤酒'
IF 性別 = '男' AND 教育程度 = '國中或高中職' AND 宗教 = '佛教' THEN '啤酒' Default rule: 啤酒
Accuracy rate on the training set: 85.92592592592592 % Accuracy rate on the test set: 85.0 %
Time taken: 5.047 s.
--- Cross Validation #3--- Cases in the training set: 540
Cases in the test set: 60 Rules: 9
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '女' AND 婚姻狀況 = '已婚' THEN '葡萄酒' IF 宗教 = '無' THEN '啤酒'
IF 職業 = '以勞力為主之員工' THEN '啤酒' IF 血型 = 'O' THEN '啤酒'
IF 性別 = '男' AND 職業 = '負責人或部門主管' THEN '啤酒' IF 性別 = '男' AND 職業 = '以腦力為主之員工' THEN '啤酒' IF 血型 = 'B' THEN '葡萄酒'
Default rule: 高粱酒
Accuracy rate on the training set: 86.11111111111111 % Accuracy rate on the test set: 85.0 %
Time taken: 5.093 s.
--- Cross Validation #4--- Cases in the training set: 540
Cases in the test set: 60 Rules: 8
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '女' AND 婚姻狀況 = '已婚' THEN '葡萄酒' IF 宗教 = '無' THEN '啤酒'
IF 職業 = '以勞力為主之員工' THEN '啤酒' IF 教育程度 = '專科或大學' THEN '啤酒'
IF 教育程度 = '國中或高中職' AND 宗教 = '道教' THEN '啤酒' IF 性別 = '男' AND 教育程度 = '國中或高中職' THEN '啤酒' Default rule: 高粱酒
Accuracy rate on the training set: 87.03703703703704 % Accuracy rate on the test set: 80.0 %
Time taken: 4.875 s.
--- Cross Validation #5--- Cases in the training set: 540
Cases in the test set: 60 Rules: 8
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '女' AND 婚姻狀況 = '已婚' THEN '葡萄酒' IF 宗教 = '無' THEN '啤酒'
IF 職業 = '以勞力為主之員工' THEN '啤酒' IF 教育程度 = '專科或大學' THEN '啤酒'
IF 教育程度 = '國中或高中職' AND 宗教 = '道教' THEN '啤酒' IF 性別 = '男' AND 教育程度 = '國中或高中職' THEN '啤酒' Default rule: 高粱酒
Accuracy rate on the training set: 87.03703703703704 % Accuracy rate on the test set: 80.0 %
Time taken: 5.907 s.
--- Cross Validation #6--- Cases in the training set: 540
Cases in the test set: 60 Rules: 8
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '女' AND 婚姻狀況 = '已婚' THEN '葡萄酒' IF 宗教 = '無' THEN '啤酒'
IF 職業 = '以勞力為主之員工' THEN '啤酒' IF 教育程度 = '專科或大學' THEN '啤酒'
IF 教育程度 = '國中或高中職' AND 宗教 = '道教' THEN '啤酒' IF 性別 = '男' AND 教育程度 = '國中或高中職' THEN '啤酒' Default rule: 高粱酒
Accuracy rate on the training set: 86.11111111111111 % Accuracy rate on the test set: 88.33333333333333 % Time taken: 5.156 s.
--- Cross Validation #7--- Cases in the training set: 540
Cases in the test set: 60 Rules: 6
IF 是否喜歡喝酒 = '不喜歡' THEN '無' IF 性別 = '男' THEN '啤酒'
IF 婚姻狀況 = '未婚' THEN '啤酒'
IF 教育程度 = '國中或高中職' AND 婚姻狀況 = '已婚' THEN '啤酒' IF 婚姻狀況 = '已婚' THEN '葡萄酒'
Default rule: 高粱酒
Accuracy rate on the training set: 86.29629629629629 % Accuracy rate on the test set: 81.66666666666667 % Time taken: 3.109 s.
--- Cross Validation #8--- Cases in the training set: 540
Cases in the test set: 60 Rules: 9
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '女' AND 婚姻狀況 = '已婚' THEN '葡萄酒' IF 教育程度 = '專科或大學' THEN '啤酒'
IF 性別 = '男' AND 年齡 = '15-19 歲' THEN '葡萄酒' IF 職業 = '以勞力為主之員工' THEN '啤酒'
IF 性別 = '男' AND 職業 = '負責人或部門主管' THEN '啤酒' IF 職業 = '以腦力為主之員工' THEN '啤酒'
IF 教育程度 = '國中或高中職' THEN '啤酒' Default rule: 高粱酒
Accuracy rate on the training set: 85.37037037037038 % Accuracy rate on the test set: 93.33333333333333 % Time taken: 5.141 s.
--- Cross Validation #9--- Cases in the training set: 540
Cases in the test set: 60
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '男' AND 婚姻狀況 = '已婚' THEN '啤酒' IF 婚姻狀況 = '未婚' THEN '啤酒'
IF 婚姻狀況 = '已婚' AND 宗教 = '佛教' THEN '葡萄酒' IF 血型 = 'A' THEN '啤酒'
IF 宗教 = '道教' THEN '葡萄酒' Default rule: 啤酒
Accuracy rate on the training set: 87.03703703703704 % Accuracy rate on the test set: 81.66666666666667 % Time taken: 3.906 s.
--- Cross Validation #10--- Cases in the training set: 540
Cases in the test set: 60 Rules: 8
IF 是否喜歡喝酒 = '不喜歡' THEN '無'
IF 性別 = '女' AND 婚姻狀況 = '已婚' THEN '葡萄酒' IF 宗教 = '無' THEN '啤酒'
IF 職業 = '以勞力為主之員工' THEN '啤酒' IF 教育程度 = '專科或大學' THEN '啤酒'
IF 教育程度 = '國中或高中職' AND 宗教 = '道教' THEN '啤酒' IF 性別 = '男' AND 教育程度 = '國中或高中職' THEN '啤酒' Default rule: 高粱酒
Accuracy rate on the training set: 85.55555555555556 % Accuracy rate on the test set: 93.33333333333333 % Time taken: 4.984 s.
--- 10-Fold Cross Validation Results ---
Accuracy Rate on Test Set | Rules Number | Conditions Number ---
85.17% +/- 1.58% | 8.1 +/- 0.31 | 10.2 +/- 0.68 Total elapsed time: 48 s.