Mind map
Data reduction by sampling
Map content
Data reduction by sampling
REASONS TO CHOOSE SAMPLING?
- • To improve runtime performance
- HOW?
- By estimating models based on a subset of the data.
- Models estimated from a sample may be as accurate as those derived from the full dataset
- (if the sample represents the population adequately)
- Models estimated from a sample may be as accurate as those derived from the full dataset
- By estimating models based on a subset of the data.
- HOW?
- For a more focused analysis
- By selecting groups of related records
- EG
- All the properties in a specific neighbourhood
- EG
- By selecting groups of related records
- • To improve runtime performance
TYPES OF SAMPLES
- Random sampling
- Stratified sampling
- Systematic or 1-in-n sampling
- For non-random (structured) samples
In IBM SPSS Modeler
- Sample node
- Used to select a subset of records for analysis,
- OR to specify a proportion of records to discard
- Sample node
Information
- Description
- Data Mining Mind Map on Data reduction by sampling, created by sue san on 09/02/2017.
- Created by
- sue san
- Views
- 1
- Rating