Property Recommendations for all Australians September 2016 Glenn - - PowerPoint PPT Presentation

Property Recommendations for all Australians

September 2016

Ben Kuai Senior Developer - Data Glenn Bunker Data Science Manager

Change the way the world experiences property Change the way the world experiences property Change the way the world experiences property Change the way the world experiences property

> 5.9 million unique audience* > Hundreds of thousands of property listings** > 10,000 real estate agencies** > 160 million consumer-property interactions**

*Source: Nielsen Digital Ratings (Monthly), July 2016 **Source: REA Internal Data (Buy and Rent), July 2016

Why build a recommendation engine?

Serendipity Serendipity Serendipity Serendipity Passive or Passive or Passive or Passive or intangible intangible intangible intangible characteristics characteristics characteristics characteristics Properties to Properties to Properties to Properties to people people people people

Implicit interest ratings

+

> > > >

No explicit No explicit No explicit No explicit functionality functionality functionality functionality required required required required More accurate More accurate More accurate More accurate ratings ratings ratings ratings Many more Many more Many more Many more implicit ratings implicit ratings implicit ratings implicit ratings

6

Collect implicit interest ratings

7

Consumer Consumer Consumer Consumer information information information information Property Property Property Property information information information information Consumer Consumer Consumer Consumer-

• property

property property property interest rating interest rating interest rating interest rating Consumer Consumer Consumer Consumer events events events events

+

Calculate implicit interest ratings

8

Collaborative Collaborative Collaborative Collaborative filtering filtering filtering filtering

Advantages of collaborative filtering

Rewards market Rewards market Rewards market Rewards market leading audience leading audience leading audience leading audience Serendipity Serendipity Serendipity Serendipity Data simplicity Data simplicity Data simplicity Data simplicity

Item-based collaborative filtering process

Item-based collaborative filtering process

Item-based collaborative filtering implementation

13

Item-based collaborative filtering Spark implementation

14

Consumer Consumer Consumer Consumer-

• property

property property property r r r rating ating ating ating RDD RDD RDD RDD Distributed Distributed Distributed Distributed row matrix row matrix row matrix row matrix Property Property Property Property-

• property

property property property similarity RDD similarity RDD similarity RDD similarity RDD Top N Top N Top N Top N prediction prediction prediction prediction

Column similarity Weighted sum

+

1. 1. 1. 1. 2. 2. 2. 2. 3. 3. 3. 3.

Item-based collaborative filtering dataflow

15

Consumer events Property information Consumer information Prediction & property similarity

Content Content Content Content-

• based

based based based filtering filtering filtering filtering

Advantages of content-based filtering

Cold Cold Cold Cold-

• start

start start start Understanding Understanding Understanding Understanding & trust & trust & trust & trust

Content profiles

Bedrooms Bedrooms Bedrooms Bedrooms Price Price Price Price Bathrooms Bathrooms Bathrooms Bathrooms Property type Property type Property type Property type Location Location Location Location

Content-based recommendations

19

• Search by property (more like this)
• Search by

Search by Search by Search by consumer consumer consumer consumer

• Consumer profile with property features
• Search property matching given consumer profile

Content-based consumer profiles

20

Consumer Consumer Consumer Consumer profile profile profile profile Consumer Consumer Consumer Consumer information information information information Property Property Property Property information information information information Consumer Consumer Consumer Consumer events events events events

Content-based dataflow

21

Consumer events Property information Elastic search Consumer profile Indexed property information

Blended recommendations

Serendipity Rewards market leading audience Data simplicity Cold-start new properties Cold-start new properties Natural understanding builds trust Differentiated to search experience

Blended recommendations

23

Collaborative filtering API Collaborative filtering API Collaborative filtering API Collaborative filtering API Content Content Content Content-

• based API

based API based API based API CF Predictions Indexed property information Property-property similarity Consumer profile Blend Blend Blend Blend API API API API

Suggested properties, and so much more… Suggested properties, and so much more… Suggested properties, and so much more… Suggested properties, and so much more…

What worked well

Sampling is fine Sampling is fine Sampling is fine Sampling is fine Know the technique Know the technique Know the technique Know the technique Keep it simple Keep it simple Keep it simple Keep it simple Iterate, test & learn Iterate, test & learn Iterate, test & learn Iterate, test & learn Subject matter Subject matter Subject matter Subject matter expertise expertise expertise expertise Bigger picture Bigger picture Bigger picture Bigger picture

Related papers

26

• Item-based Collaborative Filtering Recommendation Algorithms by Badrul Sarwar,

George Karypis, Joseph Konstan, and John Riedl GroupLens

• All-pairs similarity via DIMSUM twitter
• Accurate Methods for the Statistics of Surprise and Coincidence by Ted Dunning

Thank you Thank you Thank you Thank you

Glenn Bunker Glenn Bunker Glenn Bunker Glenn Bunker

https://au.linkedin.com/in/glenn https://au.linkedin.com/in/glenn https://au.linkedin.com/in/glenn https://au.linkedin.com/in/glenn-

• bunker

bunker bunker bunker-

• 13003112

13003112 13003112 13003112

Ben Kuai Ben Kuai Ben Kuai Ben Kuai

https://www.linkedin.com/in/ben https://www.linkedin.com/in/ben https://www.linkedin.com/in/ben https://www.linkedin.com/in/ben-

• kuai

kuai kuai kuai-

• 7b1aa73

7b1aa73 7b1aa73 7b1aa73