Postprocessing Decision Trees to Extract Actionable Knowledge

1
Post-processing Decision Treesto
ExtractActionable Knowledge

• Qiang Yang and Jie Yin
• HKUST, Hong Kong
• China
• and
• Charles X. Ling and Tielin Chen
• Department of Computer Science
• University of Western Ontario, Canada

2
CRM

• Customer Relationship Management focus on
  customer satisfaction to improve profit
• Two kinds of CRM
• Enabling CRM Infrastructure, multiple touch
  point management, data integration and
  management,
• Oracle, IBM, PeopleSoft, Siebel Systems,
• Intelligent CRM data mining and analysis,
  database marketing, customization
• Vendors/products (see later)

3
From Data Mining to Actions

• What to do to help Sammy to get loan approval?

• Action 1 (Dylan) get higher income to 80K
• Action 2 (Beatrice) get Married!!

4
Actionable vs. Passive Data Mining

• Improve customer relationship
• Actions (promotion, communication) ? changes
• What actions to take to change customers from an
  undesired status to a desired one
• From churn to loyal
• From inactive to active
• From low spending to high spending
• From non-buyers to buyers
• and still make a profit (the ultimate goal)
• Approach Post-processing Decision Trees
• Mining actions from decision trees
• Bounded action problem
• Bounded segment problem
• Our solutions

5
Post-processing Decision Trees

• Get Customer Data (marketing DB)
• Build Customer Profiles
• Search Actions for Maximal Profit
• Action Delivery

6
Step 1 Get Customer Data
Marketing DB Segmentation, data preparation,
pre-processing Define a target undesired
status and desired status
7
Step 2 Build Customer Profile on target
Automatically by Proactive Solution with
probabilities on the target
8
Step 3 Search Actions for Maximal Profit
Proactive Solution searches more desired nodes in
the profile
9
Jack , Service M, Sex M, Profit 4000
Serv M?H Rate ? ? L
Prob gain -0.1 E.Profit -400 Cost 500 E.Net
Profit -900
Prob gain 0.7 E Profit 2800 Cost ? E Net
Profit - ?
Prob gain 0.6 E Profit2400 Cost800 E
NetProfit1600
Prob gain 0.6 E.Profit2400 Cost800 E.NetProf
it1600
Prob gain 0.3 E Profit1200 Cost400 E
NetProfit800
10
Step 4 Action Deployment

• Selective deployment human intelligence,
• Customer segmentation by actions

11
Practical Issue Resource is Bounded

• Limited number of account managers
• Thus, the number of customer segments is bounded
• Research how to generate no more than K customer
  segments, such that
• for each segment, find a set of common actions to
  apply
• We call this the bounded segmentation problem
  (BSP)
• Limited number of marketing actions
• Thus, types of actions are limited
• We call this the Bounded Attribute Set Problem
  (BASP)
• Both problems are NP-hard.

12
The Bounded Segmentation Problem

• Resources are bounded!
• Group (potential) negative-class customers into
  pre-specified k customer segments.
• Recommend near optimal actions to help each of
  the k customer segments switch to a more
  profitable positive class.
• Each segment is applied by the same actions (same
  manager)
• The expected net profit is to be maximized
• Each action may have a different cost and bring
  different profits
• The Bounded Segmentation Problem is NP-Complete
• Equivalent to maximum coverage problem.
• NP-hard problem!
• We seek approximate solutions!

13
The Bounded Segmentation Problem Greedy Algorithm

• Discover who are negative-class customers.
• build decision tree as the classifier
• Group negative-class leaf nodes into k customer
  segments using greedy algorithm.
• Each customer segment ? one action set
• The total profit gain by applying such k action
  sets can be maximized.
• Algorithm is based on finding the current largest
  coverage in linear time

14
An Example K2

• If we want to find two customer segments (k2)
• It is more profitable to transform L2?L1 and
  L4?L3 than others
• Profit gain (0.9-0.2)1-0.2
  (0.8-0.5)1-0.10.7.

cost
15
Experiment on Mutual Fund Data

• GreedyBSP can find k customer segments with
  maximal profit. Result is very close to those
  found by OptimalBSP.
• GreedyBSP is more scalable than OptimalBSP.

16
Summary

• From decision-tree model building to extracting
  actions for profit
• Goal maximal net profit
• Resource is bounded
• Design optimization solutions for action
  extraction
• BASP and BSP
• Future explore more efficient solutions