CODEWITHSUNDEEP
algorithmstatisticscluster-computinginstanceweka
Aristi
Aristi

Reputation: 67

Different results of K-means algorithm in Weka

If i use any of the algorithms in Weka i have reults of the following format:

=== Stratified cross-validation ===
=== Summary ===

Correctly Classified Instances         302               63.3124 %
Incorrectly Classified Instances       175               36.6876 %
Kappa statistic                          0.3536
Mean absolute error                      0.3464
Root mean squared error                  0.4176
Relative absolute error                 85.5832 %
Root relative squared error             92.8684 %
Total Number of Instances              477     

=== Detailed Accuracy By Class ===

           TP Rate   FP Rate   Precision   Recall  F-Measure   ROC Area  Class
             0.801     0.407      0.686     0.801     0.739      0.659    1
             0.748     0.243      0.549     0.748     0.633      0.718    2
             0         0          0         0         0          0.478    3
Weighted Avg.    0.633     0.283      0.516     0.633     0.568      0.641

=== Confusion Matrix ===

     a   b   c   <-- classified as
   201  50   0 |   a = 1
    34 101   0 |   b = 2
    58  33   0 |   c = 3

But if i use k-means my results are of the following format:

=== Model and evaluation on training set ===


kMeans
======

Number of iterations: 9
Within cluster sum of squared errors: 297.46622082142716
Missing values globally replaced with mean/mode

Cluster centroids:
                            Cluster#
Attribute        Full Data         0         1         2
                     (477)     (136)     (172)     (169)
========================================================
Religion            8.6939    7.6691    8.9709    9.2367
Vote_Criterion      2.7736    2.8971    2.4942    2.9586
Sex                 1.4906    1.4559         2         1
DateBirth        1930.7652 1937.5147 1920.2965 1935.9882
Educ                3.2201    3.2721    3.2209    3.1775
Immigrant           1.6415    1.6838    1.5872    1.6627
Income              2.4675       2.5    2.5523     2.355
Occupation          3.6184    3.8162    3.2907    3.7929
Vote2013                 1         2         1         1




Time taken to build model (full training data) : 0.06 seconds

=== Model and evaluation on training set ===

    Clustered Instances

    0       136 ( 29%)
    1      172 ( 36%)
    2      169 ( 35%)

..But i want to know the correctly classified instances,the precision,the recall etc as other algorithms show me.Why is that happening and how can i make weka show me results in the first format for k-means?

Upvotes: 1

Views: 1648

Answers (2)

K. Raihana
K. Raihana

Reputation: 1

In this case, ClassificationViaClustering, a meta classifier can be used. In WEKA 3.8 it has to be downloaded through the package manager separately. Hope this helps.

Upvotes: 0

zw324
zw324

Reputation: 27180

K-Means is by itself a clustering algorithm:

Cluster analysis or clustering is the task of grouping a set of objects in such a way that objects in the same group (called cluster) are more similar (in some sense or another) to each other than to those in other groups (clusters)

so it does not have a notion of "class", thus is not used for classification (it could be made to, of course, but the performance might not be too good). Are you sure you are using it correctly here?

Also, see here (bold is mine):

You could use the meta-classifier ClassificationViaClustering in order to use the clusterers in a supervised environment.

Upvotes: 1

Related Questions