The Fifth Elephant 2014

A conference on big data and analytics

What would you recommend?

Submitted by Anand (@anandk) on Friday, 11 April 2014

videocam_off

Technical level

Intermediate

Section

Workshops

Status

Submitted

Vote on this proposal

Login to vote

Total votes:  +5

Objective

This workshop will provide the audience with a quick overview of recommendation systems & how to build one from scratch. We shall build user-user collaborative filter (CF) based recommendation engines as well as item-item CF recsys. The audience will get a flavour of the range of statistical & mathematical computations that go into a recsys.

Description

Both beginners & mildly experienced folks would have a good set of skills to take away from this session. We will together de-mystify all that one imagines about recommendation systems.

We will cover what user-user CFs are as well as item-item CFs. We will then design and build an implementation (in R) of each & apply them to standard data sets. We will go into Mahout's pre-defined algorithms & how to use them (on a single node Hadoop setup).

Speaker bio

Anand has been dabbling in systems software & application software development for the past 12 years. For the past few years his interest in data analytics has taken a lion's share of his energy. While not in front of the computer, he writes, travels, shoots (pictures), cooks (a wandering chef) and does a lot more he doesn't confess to.

Comments

  • 1
    Vinayak Hegde (@vin) 4 years ago

    The talk proposal does sound interesting. Can you give more details about the techniques and attach the slides to make a fair judgement of the talk.

Login with Twitter or Google to leave a comment