Personalization of Supermarket Product Recommendations
-
Upload
micah-love -
Category
Documents
-
view
47 -
download
0
description
Transcript of Personalization of Supermarket Product Recommendations
Julian Keenaghan 1
Personalization of Supermarket Product Recommendations
IBM Research Report (2000)R.D. Lawrence et al.
Julian Keenaghan 2
Introduction
Personalized recommender system designed to suggest new products to supermarket shoppers
Based upon their previous purchase behaviour and expected product appeal
Shoppers use PDA’s Alternative source of new ideas
Julian Keenaghan 3
Introduction continued Content-based filtering
based on what person has liked in the past measure of distance between vectors representing:
Personal preferences Products
overspecialization
Collaborative filtering items that similar people have liked Associations mining (product domain) Clustering (customer domain)
Julian Keenaghan 4
Product Taxonomy
Classes (99)
Subclasses (2302)
Products (~30000)
Fresh Beef
Petfoods …..Soft Drinks …..
Dried Cat Food
Dried Dog Food
Canned Cat Food
Friskies Liver (250g)
Beef Joints
Julian Keenaghan 5
Overview
Customer Purchase Database
Data MiningAssociations
Data MiningClustering Product
Database
MatchingAlgorithm
Cluster-specificProduct lists
Personalized Recommendation
List
Normalized
customer
vectors
Cluster
assignments
Product list
for target customer’s
cluster
Products eligible
for recommendation
Product affinities
Julian Keenaghan 6
Customer Model
Customer profileVector, C(m)
s, for each customer
At subclass level => 2303 dim spaceNormalized fractional spending
quantifies customer’s interest in subclass relative to entire customer database
value of 1 implies average level of interest in a subclass
Julian Keenaghan 7
Clustering Analysis
To identify groups of shoppers with similar spending histories
Cluster-specific list of popular products used as input to recommender
Clustered at 99-dim product-class level Neural, demographic clustering algorithms Clusters evaluated in terms of dominant attributes:
products which most distinguish members of the cluster Cluster 1 – Wines/Beers/Spirits Cluster 2 – Frozen foods Cluster 3 - Baby products, household items etc..
Julian Keenaghan 8
Associations Mining Determine relationships among product classes or subclasses Used IBM’s “Intelligent Miner for Data”
Apriori algorithm Support, Confidence, Lift factors Rule: Fresh Beef => Pork/Lamb
Support 0.016 Confidence 0.33 Lift 4.9
Rule: Baby:Disposable Nappies => Baby:Wipes
Julian Keenaghan 9
Product Model
Each product, n, represented by a 2303-dim vector P(n)
Individual entries Ps(n) reflect the “affinity” the product has
to subclass s.
Ps(n) =
0 otherwise
0.25 if C(n) C(s) (associated class)
0.5 if C(s) = C(n) (same class)
1.0 if S(n) s (associated subclass)
1.0 if s = S(n) (same subclass)
Julian Keenaghan 10
Matching Algorithm
Score each product for a specific customer and select the best matches.
Cosine coefficient metric usedC is the customer vector
P is the product vector
σ mn is the score between customer m and product n
σmn = ρn C(m). P(n) / ||C(m)|| ||P(n)||
Julian Keenaghan 11
Matching Algorithm ctd.
Limit recommendations for each customer to 1 per product subclass, and 2 per class.
10 to 20 products returned to PDA Previously bought products excluded Data from 20,000 customers Recommendations for 200
Julian Keenaghan 12
Results
Recommendations generated weekly 8 months, 200 customers from one store “Respectable” 1.8% boost in revenue from
purchases from the list of recommended products.
Accepted Recommendations from product classes new to the customer
Certain products more amenable to recommendations. Wine vs. household care. “interesting” recommendations
Julian Keenaghan 13
Summary
Product recommendation system for grocery shopping
Content and Collaborative filteringPurchasing historyAssociations MiningClustering
Revenue boosts ~2%