IMPORTANT: This is the old Data Mining forum.

I keep it online so that you can read the old messages.

Please post your new messages in the

about association rule mining

Posted by:
**
djanahanaa
**

Date: March 16, 2014 05:36AM

Hi every body

please, I have 2 question about association rule mining

1. I want to calculate the confidence of a given rule after generating frequents itemsets, we all know that the confidence=support(Y)/support(X) where X is our antecedent. My question is how can I modify your Apriori ou Close or even CHARM in order to calculate the support of a givent itemset( i need a function f we gave it an itemset and it returns its support)

2. my seconde question is about FP-tree: i have read that "FP-tree suffers from a costly storing step.(the storage compression of input dataset)" i really want to know if there is any studies that may resolv this problem.

and thanks in advance

please, I have 2 question about association rule mining

1. I want to calculate the confidence of a given rule after generating frequents itemsets, we all know that the confidence=support(Y)/support(X) where X is our antecedent. My question is how can I modify your Apriori ou Close or even CHARM in order to calculate the support of a givent itemset( i need a function f we gave it an itemset and it returns its support)

2. my seconde question is about FP-tree: i have read that "FP-tree suffers from a costly storing step.(the storage compression of input dataset)" i really want to know if there is any studies that may resolv this problem.

and thanks in advance

Posted by:
**
webmasterphilfv
**

Date: March 16, 2014 08:08AM

Hello,

(1) I would suggest to read the algorithm for association rule generation by Agrawal.

You can get information about it in this chapter:

http://www-users.cs.umn.edu/~kumar/dmbook/ch6.pdf

Section 6.2 explains Apriori for frequent itemsets.

Section 6.3 explains rule generations by using the frequent itemsets.

I would suggest to read this first before considering CHARM and Close because these algorithms find a subset of all frequent itemsets. They are generally used to find a subsets of all association rules such as closed association rules...

(2)FPGrowth works in two steps. It first compress the database in a FPTree structure. Then it uses the FPTree to find frequent itemsets. The cost of building the FPTree is in my opinion very low because FPGrowth just need to read the database once to build the FP-Tree. Actually, the second steps should be more costly in my opinion.

(1) I would suggest to read the algorithm for association rule generation by Agrawal.

You can get information about it in this chapter:

http://www-users.cs.umn.edu/~kumar/dmbook/ch6.pdf

Section 6.2 explains Apriori for frequent itemsets.

Section 6.3 explains rule generations by using the frequent itemsets.

I would suggest to read this first before considering CHARM and Close because these algorithms find a subset of all frequent itemsets. They are generally used to find a subsets of all association rules such as closed association rules...

(2)FPGrowth works in two steps. It first compress the database in a FPTree structure. Then it uses the FPTree to find frequent itemsets. The cost of building the FPTree is in my opinion very low because FPGrowth just need to read the database once to build the FP-Tree. Actually, the second steps should be more costly in my opinion.

Posted by:
**
djanahanaa
**

Date: March 18, 2014 01:45PM

Thanks for this relpy

i will read the document to understand association rule mining

for FP-tree I have read a paper where the authors propose a modified structure of FP-tree in order to avoid the problem of huge size of database

thanks again for the anwser

i will read the document to understand association rule mining

for FP-tree I have read a paper where the authors propose a modified structure of FP-tree in order to avoid the problem of huge size of database

thanks again for the anwser

Posted by:
**
khairy
**

Date: March 21, 2014 11:46AM

Dear Sir

suppose we have the following transactions

100 bread cheese eggs juice

200 bread cheese juice

300 bread milk yoghurt

400 bread juice milk

500 cheese juice milk

if the support is 50% and the confidence is 75%

the 1st frequent itemsets will be

bead 4

cheese 3

juice 4

milk 3

and the 2nd candidate itemsets will be

bread cheese 2

bread juice 3

bread milk 2

cheese juice 3

cheese milk 1

juice milk 2

please i need a help to discover the frequent

2-itemset

3-itemset

.........

.........

and finally the association rule from the generated frequent itemsets

thanks in advance

Edited 1 time(s). Last edit at 03/21/2014 11:48AM by khairy.

suppose we have the following transactions

100 bread cheese eggs juice

200 bread cheese juice

300 bread milk yoghurt

400 bread juice milk

500 cheese juice milk

if the support is 50% and the confidence is 75%

the 1st frequent itemsets will be

bead 4

cheese 3

juice 4

milk 3

and the 2nd candidate itemsets will be

bread cheese 2

bread juice 3

bread milk 2

cheese juice 3

cheese milk 1

juice milk 2

please i need a help to discover the frequent

2-itemset

3-itemset

.........

.........

and finally the association rule from the generated frequent itemsets

thanks in advance

Edited 1 time(s). Last edit at 03/21/2014 11:48AM by khairy.

Posted by:
**
webmasterphilfv
**

Date: March 21, 2014 06:06PM

I may check your result. But you should first try to calculate it by yourself because it looks like an homework.