Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

"Cluster Valid"

sieniek

how to implement cluster Valid like Davies-Bouldin’s Index or Classification Entropy

Find more posts tagged with

AI Studio

Clustering

Accepted answers

All comments

land

Hi,
here's a sample process:

<operator name="Root" class="Process" expanded="yes">
    <description text="This process plots the learning curve, i.e. the performance with respect to the number of examples which is used for learning."/>
    <parameter key="logverbosity"	value="warning"/>
    <parameter key="random_seed"	value="2004"/>
    <operator name="Iris" class="ArffExampleSource">
        <parameter key="data_file"	value="C:\rm_workspace\sample\data\iris.arff"/>
        <parameter key="label_attribute"	value="class"/>
    </operator>
    <operator name="KMeans" class="KMeans">
    </operator>
    <operator name="ClusterCentroidEvaluator" class="ClusterCentroidEvaluator">
        <parameter key="main_criterion"	value="Davies Bouldin"/>
    </operator>
</operator>

Greetings,
Sebastian

sieniek

Czy istnieją inne wdrożone metody oceny jakości grupowania

land

Hi,
please write in english, so that all users can read your posts.

There are some other operators in the group Validation/performance/Clustering available. You might read their description in order to check if they suit your need.
One additional method, which is only applicable when you already know what would be the correct clustering, is the Cluster2Prediction operator, which will match the cluster attribute to the labels and transforms it to a prediction attribute, so that you can use the Classification Performance measures in order to estimate the goodness of clustering.

Greetings,
Sebastian

sieniek

sorry
Are there other methods to assess the quality of implemented clustering

land

Hi,
since you cannot really assess the quality of a clustering, there's not too much you can do. But everything what's included is in the same group as the operator I suggested. Try it out and feel free to invent another measure operator for clustering.

Greetings,
Sebastian

sieniek

hello
and Dunn's index is implemented in the program? if not can you do this yourself and how?

land

Hi,
you would have to either implement an own operator and put it inside a plugin, or you could take some money and pay us to do so

For implementing your own plugin, there are many posts inside this forum you could search and which will help you if you hang.

Greetings,
Sebastian