An introduction to Apache Mahout

1
Apache Mahout

• What is it ?
• How does it work ?
• Machine Learning
• Algorithms
• Install

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
2
Mahout What is it ?

• Machine learning
• For large data
• Based on Hadoop
• But can work on a non Hadoop cluster
• Scaleable
• Licensed by Apache

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
3
Mahout How does it work ?

• Uses Hadoop Map Reduce
• Has many supplied algorithms
• Supports four use cases
• Recommendation mining
• Clustering
• Classification
• Frequent Itemset Mining

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
4
Mahout - Machine Learning

• Machine learning what does it mean ?
• A branch of artificial intelligence
• Systems that learn from data
• Classify data after learning
• Learn on test data sets
• Generalisation the ability to classify unseen
  data sets
• after learning

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
5
Mahout Algorithms

• Some of the available algorithms (among many
  others)?
• Collaborative filtering
• Narrow Sense make predictions about user
  interests by collecting preferences
• General - Multi agent collaboration for
  information filtering
• Mean shift clustering
• Mode seeking, used for visual tracking
• Parallel frequent pattern mining
• Find unique features

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
6
Mahout Install

• So how do we install Mahout and test it ?
• Install Maven
• sudo apt-get install maven3
• Install Apache Mahout
• You will need subversion installed
• svn co http//svn.apache.org/repos/asf/mahout/trun
  k
• Go to dir containing pom.xml file
• mvn install in ./trunk
• Full details available in the Mahout install
  guide on our web site shop

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
7
Mahout Test Install

• So let us run a test
• cd MAHOUT_HOME/examples/bin
• ./build-reuters.sh
• choose option 1 kmeans clustering
• Should finish with see next slide
• Full details available in the Mahout install
  guide on our web site shop

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
8
Mahout Test Install

• cd MAHOUT_HOME/examples/bin
  ./build-reuters.sh
• Please call cluster-reuters.sh directly next
  time. This file is going away.
• Please select a number to choose the
  corresponding clustering algorithm
• 1. kmeans clustering
• 2. fuzzykmeans clustering
• 3. lda clustering
• Enter your choice 1
• ok. You chose 1 and we'll use kmeans Clustering
• .................................
• Inter-Cluster Density NaN
• Intra-Cluster Density 0.0
• CDbw Inter-Cluster Density NaN
• CDbw Intra-Cluster Density NaN
• CDbw Separation NaN
• Full details available in the Mahout install
  guide on our web site shop

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
9
Contact Us

• Feel free to contact us at
• www.semtech-solutions.co.nz
• info_at_semtech-solutions.co.nz
• We offer IT project consultancy
• We are happy to hear about your problems
• You can just pay for those hours that you need
• To solve your problems