Subject: | |
From: | |
Reply To: | |
Date: | Fri, 19 Jul 2013 09:25:13 -0400 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
On Jul 17, 2013, at 7:24 PM, Hooman HOMAYOUNFARD wrote:
> I am going to read a csv file including of the transactions for number of items and
> find the frequent itemsets out using a genetic algorithm approach.
> Well, as I found in the tutorials ECJ creates number of genomes and evaluate it
> by an individual to find how the genome(s) is(are) fitted to the target individual.
>
> There are number of questions:
>
> 1- Is it possible to have number of individuals?
Of course, this is the default. Hooman, this question suggests that you are not very familiar with evolutionary computation methods. This list is not for answering EC questions, but specifically discussing ECJ's implementation.
> 2- if yes, then: How do the genomes should be evaluated with a number of
> transactions (individuals) of the itemsets?
ECJ is not a frequent itemset tool. It is up to you to decide how to encode such things in your individuals.
> 3- How I can change the output to print the most fitted genome in each
> generation.
Try setting stat.do-generation=true in your parameters. See Section 3.7 of the manual.
Sean
|
|
|