Chapter 1: Introduction to Predictive Modeling 1.1 Applications 1.2 Generalization 1.3 JMP...

Preview:

Citation preview

Chapter 1: Introduction to Predictive Modeling

1.1 Applications

1.2 Generalization

1.3 JMP Predictive Modeling Platforms

Chapter 1: Introduction to Predictive Modeling

1.1 Applications1.1 Applications

1.2 Generalization

1.3 JMP Predictive Modeling Platforms

Objectives Describe common applications of predictive modeling

in business, science, and engineering. Describe typical data that is available for predictive

modeling. Define commonly used terms used in predictive

modeling.

3

Predictive Modeling Applications

4

Database marketing

Financial risk management

Fraud detection

Process monitoring

Pattern detection

Healthcare Informatics

The Data

5

Experimental Opportunistic

Purpose Research Operational

Value Scientific Commercial

Generation Actively controlled Passively observed

Size Small Massive

Hygiene Clean Dirty

State Static Dynamic

inputs target

Predictive Modeling Data

6

Training Data

Training data case: categorical or numeric input and target measurements

Types of Targets Supervised Classification

– Event/no event (binary target)– Class label (multiclass problem)

Regression– Continuous outcome

7

Continuous Targets Healthcare Outcomes

– Target = hospital length of stay, hospital cost Liquidity Management

– Target = amount of money at an ATM machine or in a branch vault

Process Volatility– Target = moving range of yields

Sales– Target = dollar value of sales

8

Measurement LevelsThree types in JMP Continuous Ordinal Nominal

JMP automatically performs specific types of analyses based on the measurement level of the target. For example, linear regression versus logistic regression.

In some platforms, ordinal and nominal variables inputs are handled differently.

9

Chapter 1: Introduction to Predictive Modeling

1.1 Applications

1.2 Generalization1.2 Generalization

1.3 JMP Predictive Modeling Platforms

Objectives Define generalization. Define honest assessment. Describe how honest assessment can be done in JMP.

11

The Scope of Generalization Model Selection and Comparison

– Which model gives the best prediction? Decision/Allocation Rule

– What actions should be taken on new cases? Deployment

– How can the predictions be applied to new cases?

12

Model Complexity

13 ...

Model Complexity

14

Not complex enough

...

Model Complexity

15

Too complex

Not complex enough

Honest Assessment: Data Splitting

16

Data Partitioning

17

Training Data

inputs target

...

Data Partitioning

18

Training Data Validation Data

inputs target inputs target

...

Data Partitioning

19

Training Data Validation Data

Partition available data intotraining and validation sets.

inputs target inputs target

5

4

3

2

1

Predictive Model Sequence

20

Create a sequence of models with increasing complexity.

ModelComplexity

Training Data Validation Data

inputs target inputs target

Model Performance Assessment

21

ValidationAssessment

Rate model performance using validation data.

Training Data Validation Data

inputs target inputs target

5

4

3

2

1

ModelComplexity

3

Model Selection

22

ModelComplexity

ValidationAssessment

Select the simplest model with the highest validation assessment.

Training Data Validation Data

inputs target inputs target

Chapter 1: Introduction to Predictive Modeling

1.1 Applications

1.2 Generalization

1.3 JMP Predictive Modeling Platforms1.3 JMP Predictive Modeling Platforms

Objectives Show the platforms that will used in the class.

24

Accessing the Neural or Partition Platforms

25

Partition Platform Dialog

26

Neural Platform Dialog

27

Recommended