Generative Models for Customer Data

1
Generative Models for Customer Data

• Elena Eneva
• Technically Speaking
• 12 Nov 2001

2
Data Mining

• the non-trivial extraction of implicit,
  previously unknown, and potentially useful
  knowledge from data for decision support,
  prediction, forecasting, and estimation.

3
The project at a glance

• Data Analysis Model Synthetic
  Testing

Data
4
Data

• N-to-M relationships
• Products, Customers (anonymous retailer)
• People, Web Pages (MSNBC users)
• Time, Access Location (HP disk traces)
• Bi-partite graphs
• What we are really dealing with Matrices

5
Analysis

• Association rules
• Distributions

Models

• Model bank
• Generator

Synthetic Data

• Anonymized realistic data

6
Testing

• Marginal distributions
• Customers
• Products
• Mutual entropy
• Scree Plot (eigenvalues)

7
Different Uses

• The generated rules
• The generated data
• (helping the data-poor people)
• For those who build the tools to find patterns
• For those who use the tools to find patterns

8
Questions?
Ideas?
The end.
9
Scree Plot

• two dimensional graph with factors on the x-axis
  and eigenvalues on the y-axis. Eigenvalues are
  produced by a process called principal components
  analysis (PCA) and represent the variance
  accounted for by each underlying factor.

10
Synthetic Data

• Syn data

11
Models

• Model bank