Affinity Analysis for Selecting "Next Best Activity"

1
Affinity Analysis for Selecting Next best
Activity
Tom Breur London, 26 September 2007 tombreur_at_xlntc
onsulting.com www.xlntconsulting.com 31-6-463
468 75
2
Agenda

• Affinity analysis what is it?
• Uni-directional affinity analysis
• How does it work?
• One-directional affinity analysis
• Different types of scaling
• An example case application of scaling

3
Affinity analysis what is it?

• Products and services customers choose tell a lot
  about them, and their preferences
• Alikeness of customers is determined by their
  geometric proximity in hyperspace
• Many different distance functions possible
• Algorithm sometimes also referred to as Case
  Based Reasoning
• Affinity Analysis ? Collaborative Filtering

4
?
?
?
?
5
(No Transcript)
6
?
?
7
?
?
8
Uni-directional affinity analysis
v4
v2
v3
v12
v8
v6
v1
v5
v7
v9
v10
v11
p1
0.2
0.1
0.3
0.4
0.8
0.7
0.5
0.1
0.1
0.9
0.9
0.7
p2
0.2
0.8
0.2
0.1
0.6
0.3
0.4
0.8
0.2
0.3
0.8
0.7
p3
0.2
0.1
0.3
0.3
0.4
0.4
0.8
0.7
0.7
0.7
0.3
0.8
p4
0.2
0.3
0.5
0.3
0.1
0.4
0.4
0.8
0.3
0.5
0.9
0.7
p5
0.4
0.3
0.8
0.3
0.7
0.2
0.8
0.5
0.2
0.5
0.7
0.9
p6
0.5
0.2
0.1
0.8
0.8
0.9
0.7
0.3
0.4
0.5
0.3
0.9
9
Uni-directional affinity analysis
10
Uni-directional affinity analysis
11
Uni-directional affinity analysis
12
One-directional affinity analysis

• Guttman scaling assumes perfect patterns
• Rasch Scaling multiplicative model
• Mokken Scaling non-parametric model

13
An application of scaling (1)
lump sum policy
consumer credit
internet banking
credit card
savings account
revolving credit
current account
14
An application of scaling (2)
mutual funds (other)
mutual funds (risk averse)
premium savings
credit card
savings account
revolving credit
current account
15
Conclusion

• Uni-directional affinity analysis
• Flexible with regards to optimal recommendation
• Technical solution is computationally efficient
• One-directional affinity analysis (scaling)
• Provides insight in different needs segments
• Useful as a prediction
• Possible input as predictor variables

morefromtom.com