Tiểu luận Khai phá dữ liệu: Sử dụng phần mềm weka khai phá bộ dữ liệu meta data

1.Mô tả bài toán Meta Data được sử dụng để đưa ra lời khuyên về phương pháp phân loại thích hợp cho một số liệu cụ thể (lấy từ các kết quả của Statlog dự án) 2.Xây dựng cơ sở dữ liệu -Dataset sửdụng: Meta Data -Thông tin dataset -Các thuộc tính: DS_Name{Aust_Credit,BT,Belgian,CUT,Chromosone,Credit,DNA,Diabetes,Digits,Faults,German_Credit,Head,Heart,KlDigits,Letters,NewBelgian,SatImage,Segment,Shuttle,Technical,TseTse,Vehicle} T real N real p real k real Bin real Cost real SDratio real correl real cancor1 real cancor2 real fract1 real fract2 real skewness real kurtosis real Hc real Hx real MCx real EnAtr real NSRatio real Alg_Name {Ac2,Alloc80,BackProp,Bayes,BayesTree,C4.5,CART,Cal5,Cascade,Castle,Cn2,Default,Dipol92,Discrim,ITrule,IndCART,KNN,Kohonen,LVQ,LogDisc,NewId,QuaDisc,RBF,Smart} Norm_error real -Training data 3.TIến hành trong Weka -Đưa dữ liệu vào Weka

BTL KHAI PHA DU LIEU

Dé bai :Meta Data

Sinh viên thực hiện: Vũ Lê Hồng

Lớp : HTTT6 GVHD :H6 Nhat Quang

1.Mơ tả bài tốn

Meta Data được sử dụng để đưa ra lời khuyên về phương pháp phân loại thích hợp cho một số liệu cụ thể (lấy từ các kết quả của Statlò dự

án)

2.Xây dựng cơ sở dữ liệu

Meta Data 2011

-Dataset sudung: Meta Data

- Thông tin dataset

Data Set Characteristics: || Multivariate Number of Instances: || 528 || Area: N/A Attribute Characteristics: || Categorical, Integer, Real || Number of Attributes: || 22 || Date Donated 1996-03-01 Associated Tasks: | Classification Missing Values? 10937

-Các thuộc tính:

DS Name{Aust_Credit,BT,Belgian,CUT,Chromosone,Credit,DNA,Diabetes,Digits,Faults,G

erman Credit,Head,Heart,KlDigits, Letters ,NewBelgian,SatImage, Segment, Shuttle, Tech

nical, TseTse, Vehicle}

T real

N real

p real

k real

Bin real

Cost real

SDratio real

correl real

cancorl real

cancor2 real

fractl real

fract2 real

skewness real

kurtosis real

Hc real

Hx real

MCx real

EnAtr real

NSRatio real

Alg_ Name

{Ac2,Alloc80,BackProp, Bayes, BayesTree,C4.5,CART, Cal5, Cascade, Castle, Cn2,Default,D 1poL92,D1scrim, TTruLe, TndCART ,KNN, Kohonen, LVQ,LogD1sc, NewTd, QuaD1sc, RBF , Smart}

Norm_error real

-Training data

3.TIến hành trong Weka

-Dua dt liéu vao Weka

Í # :

Preprocess | Classify | Cluster | Associate | Select attributes | Visualize |

| Openafie |[ OpenURL || OpenDB || Geneate | Undo J{ Edit l Save |

Filter

[ Choose |None Í Apply | Current relation Selected attribute

Relation: meta Name: T Type: Numeric

Instances: 528 Attributes: 22 Missing: 0 (0%) Distinct: 20 Unique: 0 (0%)

Minimum 270 Mean 4569.045

=|| | Class: Norm_error (Num) >| Visualize All |

360

48

24 24 24 24 24

CT TL]

-Sử dụng toàn bộ dữ liệu dé training

Meta Data 2011

Preprocess | Classify | Cluster | Associate | Select attributes | Visualize |

Classifier

Test options Classifier output

© Use training set kurtosis ˆ

Hc

© Supplied test set Set Hx

@ Cross-validation Folds 10 | MCx

= > ] EnAtr

© Percentage split ⁄ |66 | NSRatio

[ More options | Alg_Name

Norm_error Test mode:10-fold cross-validation (Num) Norm_error xv

— === Classifier model (full training set) ===

Result list (right-click for options) ZeroR predicts class value: 99.55247727272732 E

07:13:56 - rules.ZeroR

Time taken to build model: Oseconds

=== Cross-validation ===

=== Summary ===

Mean absolute error 151.5401 Root mean squared error 764.994 Relative absolute error 100 $ Root relative squared error 100 $ Total Number of Instances 528

Status

-Nội dung kếtquả

=== Classifier model (full training set) ===

ZeroR predicts class value: 99.55247727272732

Time taken to build model: Oseconds

=== Cross-validation ===

Correlation coefficient

Mean absolute error

Root mean squared error

Relative absolute error

Root relative squared error

Total Number of Instances

-Cay quyét dinh

-0.1108 151.5401 764.994

100 %

100 %

528

|| Weka Classifier Tree Visualizer: 16:38:21 - treesJ48 (acute) L=i

Tree View

mu

= no

meno

`

-05% dữ liệu để xây dựng, 5% dữ liệu test

5

Meta Data 2011

lợi

Fereprocess| Cz=5y [cher | Associate | Sclectatibutes [use]

Classifier

| Choose |ZeroR

Test options

© Use training set

Hc

© Supplied test set Set Hx

© Cross-validation Folds MCx

EnAtr

@ Percentage spit % NSRatio

[ More options | Alg_Name

Norm_error

| Test mode:split 95.0% train, remainder test

vr

| (Num) Norm_error

Result list (right-click for options) ZeroR predicts class value: 99.55247727272732

07:13:56 - rules.ZeroR

07:17:57 -rules.ZeroR Time taken to build model: Oseconds

=== Evaluation on test split ===

Correlation coefficient 0 Mean absolute error 217.8454 Root mean squared error 603.4871 Relative absolute error 100 $ Root relative squared error 100 $ Total Number of Instances 26

-90% dữliệuđêxâydựng, 10% dữliệu test

-85% dữliệuđêxâydựng, 15% dữ liệu test

“® Weka Explorer

Classifier

I|[ Preprocess| Classify | Cluster | Associate | Select attributes | visualize |

| Choose |348-C0.25-m2

Test options Classifier output

© Use training set

E

- Size of the tree : 7

© Supplied testset | = Set |

© Cross-validation Folds

@ Percentage split % |35 Time taken to build model: 0 seconds

More options === Evaluation on test split ===

=== Summary ===

| (Nom) Viém th?n v |

Correctly Classified Instances 18 100 $

| Start [ Stop Incorrectly Classified Instances 08 08 $

Kappa statistic 1

Result list (right-dick for options) Mean absolute error 0

16:38:21 - trees.J48 Root mean squared error 0

16:42:26 - trees J48 Relative absolute error 0

16:46:31 - trees.J48 Root relative squared error 0

5 : Total Number of Instances 18

=== Detailed Accuracy By Class ===

TP Rate FP Rate Precision Recall F-Measure ROC Area Class

Weighted Avg 1 0 1 1 1

=== Confusion Matrix ===

a b < classified a3

Status

-70% dữliệuđêxâydựng, 30% dữliệu test

Meta Data

-

® Weka Explorer

Preprocess | Classify | Cluster | Associate | Select attributes | Visualize |

Classifier

Choose | 348 -C.0.25-M2

Test options Classifier output

© Use training set Number of Leaves : 4

© Cross-validation Folds

@ Per ge split % m ` Time taken to build model: 0 seconds

Size of the tree : 7

=== Evaluation on test split ===

| ion) vim thn " === Summary ===

[stat] Stop Correctly Classified Instances 36 100 $

Incorrectly Classified Instances 0 0 $

Result list (right-click for options) Kappa statistic 1

16:38:21 - trees.148 Mean absolute error 0

16:42:26 - trees 148 Root mean squared error 0

16:46:31 - trees.J48 Relative absolute error 0 $

16:47:52 - trees 148 Root relative squared error 0 $

16:48:21 - trees.J48 Total Number of Instances 36

=== Detailed Accuracy By Class ===

TP Rate FP Rate Precision Recall F-Measure ROC Area Class

Weighted Ava 1 0 1 1 1 1

=== Confusion Matrix ===

| a b- < classified as

24 0 | a=ne

Status

-50% dữliệuđêxâydựng, 50% dữliệu test

Classifier

Test options Classifier output

Size of the tree : 7

© Supoed test et

© Cross-validation Folds [10 |

Time taken to build model: 0.01 seconds

@ Percentage split %

More options === Evaluation on test split ===

=== Summary ===

| ion) vim tn v _

Correctly Classified Instances 60 100 $ [ Start ] Stop | Incorrectly Classified Instances 0 0 $

Kappa statistic 1

Result list (right-click for options) Mean absolute error 0

16:38:21 - trees 148 Root mean squared error 0

16:42:26 - trees.J48 Relative absolute error 0

16:46:31 - trees.148 Root relative squared error 0

16:47:52 - trees.148 Total Number of Instances 60

=== Detailed Accuracy By Class ===

TP Rate FP Rate Precision

1 0 1

1 0 1 Weighted Avg 1 0 1

=== Confusion Matrix ===

a b < classified as

39 0| a=no

0 21 | b = yes

Recall F-Measure

1 1

1 1

1 1

ROC Area Class

no yes

Status