The Data Mining Forum 
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 new forum: https://forum2.philippe-fournier-viger.com/index.php
TopKClassRule Running error - Java heap space - out of memory
Posted by:
Diep Dao
Date: January 02, 2019 03:33PM
Hi Philippe,
I am running the TopKClassRule algorithm on a dataset to mine class association rules.
I am getting an error message
"ERROR MESSAGE = java.lang.OutOfMemoryError:Java heap space"
when I aim to output a large set of rules or increase the mincof value. With the current dataset, it seems to work only for k = 10 and mincof =0.1.
Do you know if there is a solution to this?
Many thanks,
Diep
I am running the TopKClassRule algorithm on a dataset to mine class association rules.
I am getting an error message
"ERROR MESSAGE = java.lang.OutOfMemoryError:Java heap space"
when I aim to output a large set of rules or increase the mincof value. With the current dataset, it seems to work only for k = 10 and mincof =0.1.
Do you know if there is a solution to this?
Many thanks,
Diep
Re: TopKClassRule Running error - Java heap space - out of memory
Posted by:
Diep Dao
Date: January 03, 2019 04:32AM
I just wanted to add that I have tried the -Xms and -Xmx with maximum 2gb memory setup when running the application.
Re: TopKClassRule Running error - Java heap space - out of memory
Posted by:
webmasterphilfv
Date: January 06, 2019 04:06AM
Hi,
I see. It means that the search space is perhaps too big. There are a few solutions:
- You can use the optional parameter "maximum antecedent size. If you set it to a small value, it will greatly reduce the search space and the algorithm will consume less memory.
- You can also use the optional parameter to set some required items in the consequen of rules. This will also reduce the number of possibilities that must be considered by the algorithm. It will reduce the memory and the runtime.
- You could do some preprocessing on your data to remove some information that is not useful. Or you could use some part of your data rather than all the data. For example, if you have 1,000,000 transactions, you could perhaps use the first 10,000 transactions for testing, or you could preprocess the data to remove some information that you don't need or transform it.
That is my main suggestions.
Thanks for using SPMF!
Best regards,
Philippe
I see. It means that the search space is perhaps too big. There are a few solutions:
- You can use the optional parameter "maximum antecedent size. If you set it to a small value, it will greatly reduce the search space and the algorithm will consume less memory.
- You can also use the optional parameter to set some required items in the consequen of rules. This will also reduce the number of possibilities that must be considered by the algorithm. It will reduce the memory and the runtime.
- You could do some preprocessing on your data to remove some information that is not useful. Or you could use some part of your data rather than all the data. For example, if you have 1,000,000 transactions, you could perhaps use the first 10,000 transactions for testing, or you could preprocess the data to remove some information that you don't need or transform it.
That is my main suggestions.
Thanks for using SPMF!
Best regards,
Philippe