ClusterEnG Application: DBSCAN

Loading...

Do curso por University of Illinois at Urbana-Champaign

Cluster Analysis in Data Mining

142 classificações

University of Illinois at Urbana-Champaign

Cluster Analysis in Data Mining

142 classificações

Curso 5 de 6 em Specialization Data Mining

Discover the basic concepts of cluster analysis, and then study a set of typical clustering methodologies, algorithms, and applications. This includes partitioning methods such as k-means, hierarchical methods such as BIRCH, and density-based methods such as DBSCAN/OPTICS. Moreover, learn methods for clustering validation and evaluation of clustering quality. Finally, see examples of cluster analysis in applications.

Na lição

Week 2

ClusterEnG Overview5:01

ClusterEnG: K-Means and K-Medoids3:18

ClusterEnG Application: AGNES4:04

ClusterEnG Application: DBSCAN2:48

Conheça os instrutores

Jiawei Han
Abel Bliss Professor
Department of Computer Science

Hi, my name is Sun lee Jong.

I'm a kitchen assistant of this Coursera course.

Today, I'm going to introduce, how to use an online cresting tool to apply and

to visualize the DBSCAN algorithm from the real data set.

DBSCAN algorithm is introduced in test the base algorithm.

If you're not familiar with the technical details you may want to review the slides.

The data site that we're using today is a Yelp restaurant data site.

It contains locations of a hundred restaurants in Waterloo, Canada.

Each location is represented by its longitude and latitude.

And we find this data set in course material.

To start with, we upload a data set file to the website,

select the given data set file and then click on Start button.

The website is going to have it automatically uploaded.

Then, make sure we're clustering by rows, instead of clustering by columns.

Because each row here actually represents a location.

Now, we choose a classroom method to the DBSCAN and

insert the parameter value for epsilon.

Epsilon is the parameter in DBSCAN algorithm, defines the neighborhood.

Here, we choose epsilon to be 0.005.

Click on Submit.

Here is a visualization of the clustering results.

Because we are running the algorithm on a two dimensional dataset.

We only need to look at this very first visualization figure.

Each card represents a certain cluster.

And one of our cluster actually represents the others.

As you can see from this visualization, one of the most important characteristics

of the DBSCAN algorithm, is that they can generate clusters of arbitrary shape,

in self clusters of just.

So I want to try different parameter settings

to see how the classroom result is different.

And I also want to try the classroom algorithms different view of datasets.

If you're interested about exploring more about DBSCAN algorithms,

here are some interesting open questions about the algorithm.

Thank you.

[SOUND]

[MUSIC]

[SOUND]

Explore nosso catálogo

Registre-se gratuitamente e obtenha recomendações, atualizações e ofertas personalizadas.

O Coursera proporciona acesso universal à melhor educação do mundo fazendo parcerias com as melhores universidades e organizações para oferecer cursos on-line.

© 2018 Coursera Inc. Todos os direitos reservados.

Baixar na App Store

Baixar no Google Play