Skip to content

review3.md

Latest commit

 

History

History
45 lines (29 loc) · 1.73 KB

review3.md

File metadata and controls

45 lines (29 loc) · 1.73 KB

home | copyright ©2016, tim@menzies.us

overview | syllabus | src | submit | chat

Review 3

Spatial Trees

Spatial tree use some measure of confusion M to split S the data, the recurse on each split till some halting H criteria is meet, after which some condensation operator C is applied to the summarize the leaf splits. Describe M,S,H,C for

  • J48 (aka c45). Hint: explores discrete classes
  • Cart. Hint: sometimes explores continuous classes.
  • Recusrive k=2 means. Hint: ignores classes
  • NB Tree. Hint: you've never heard of it before. It builds Naive Bayes classifiers for each leaf.
  • X Tree. Hint: this is a generalization of NB tree
  • PDDP: Hint. PCA:

BTW, for more on spatial trees, see [http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.226.5060]

Random Forests

What is the difference between a spatial tree learner and a random forest?

Explain why a random forest can better handle larger data sets with more variance than a spatial tree learner.

Evaluation Criteria

Make a case that precision is stupid.

Make a case that accuracy is stupid.

Make a case that recall is stupid.

What is AUC(effort, recall)? Hint.