Data reduction

1
Data Reduction

• data reduction

2
Conti..

• -data reduction-reduced representation of data
• in volume but produce the same result
• gtdb./dwh may store terabytes of data, complex
  data
• analysis may take more time
• -Data reduction strategies
• 1Dimensionality limitation-
• gtselect best attribute or remove unimportant
  one.

3
Cont

• 2Numerosity reduction-
• gtreduce data volume by taking smaller data
  
• representation
• 3Data compression-
• gtreduce size of large file fast to transfer
  
• over
• a network/internet
• 1
• -curse of dimensionality-

4
Cont..

• dimensionality increase data also increasingly
  
• sparse
• -Dimensionality reduction-
• allow easier visualization
• reduce time and space during mining
• eliminate irrelevant feature and reduce
  noise
• -Method- attribut subset selection
• one way to reduce dimensionality is selection
  
• best attribute.

5
Cont.

• -avoid redundant attribute
• -help to avoid irrelevant attribute stud id in
  GPA
• -Heuristic search in attribute selection
• best stepwise feature selection
• the next attribute condition to the first
• process continue until performance of combined
  attribute start to decline.