Results - Rules and Performance

Intro ] 2002-06-26-even-better-rules ] 2002-07-27-66 ]
2002-07-29-99 ] 2003-03-03-random-test ] Test-Matrix.xls ]
compare-duplicate-lookup-vector-tree-1 ] compare-previous-dups-no-previous-1 ] [ purpose ]
reduced-k=1-2002-05-09 ] trace-2002-05-03-16:53 ] very first rules ever ]

caches ] [ old-incomplete ] [ performance-improvements ] [ promoters ] [ sleep ] [ stock-market ]

The purpose of these experiments is to answer the following:

How much faster is using previous duplicate hash?

How many candidates get pruned using modified Apriori Prune check subsets?

How does check subsets compare in speed to count subsets in the
ARMiner hash tree?

Does post candidate generation garbage collection free much memory?

Does post candidate generation garbage collection make mining faster
or slower?

How much faster is using remove candidates?

What percentage of instance-itemset comparisons does removing
candidates save?

How many failed candidate generation joins are avoided using prune
candidate generation?
 

by: Keith A. Pray
Last Modified: July 4, 2004 8:03 AM
© 2004 - 1975 Keith A. Pray.
All rights reserved.