【QUESTION】Any way to limit the minimum length of each pattern? about seq2pat HOT 6 CLOSED

Thanks @Wwwwei for your interests. To your question, I don't see a way by applying currently available constraints to limit the minimum length of patterns. The span constraint you used above is actually requiring MAX(att_order) - MIN(att_order) of a pattern to be within the range [5, 10], which doesn't necessarily limit the minimum length of pattern. Besides, it may not find patterns satisfying the constraint since the maximum span of pattern would be 3, that is falling outside of the specified range.

The easiest way for now would be a post-processing step. What we shall do is to get patterns without limiting the minimum length first, then we select the patterns that can pass the desired requirement. I hope this can help to solve the problem.

from seq2pat.

yes
good question

from seq2pat.

Thanks @Wwwwei for your interests. To your question, I don't see a way by applying currently available constraints to limit the minimum length of patterns. The span constraint you used above is actually requiring MAX(att_order) - MIN(att_order) of a pattern to be within the range [5, 10], which doesn't necessarily limit the minimum length of pattern. Besides, it may not find patterns satisfying the constraint since the maximum span of pattern would be 3, that is falling outside of the specified range.

The easiest way for now would be a post-processing step. What we shall do is to get patterns without limiting the minimum length first, then we select the patterns that can pass the desired requirement. I hope this can help to solve the problem.

Thanks for your prompt reply. I will try post-processing.
By the way, is there any way to speed up and reduce memory? I am suffering from high time and memory consumption in a big dataset.
Thank you very much.

from seq2pat.

Cool

from seq2pat.

Thanks @Wwwwei for your interests. To your question, I don't see a way by applying currently available constraints to limit the minimum length of patterns. The span constraint you used above is actually requiring MAX(att_order) - MIN(att_order) of a pattern to be within the range [5, 10], which doesn't necessarily limit the minimum length of pattern. Besides, it may not find patterns satisfying the constraint since the maximum span of pattern would be 3, that is falling outside of the specified range.
The easiest way for now would be a post-processing step. What we shall do is to get patterns without limiting the minimum length first, then we select the patterns that can pass the desired requirement. I hope this can help to solve the problem.

Thanks for your prompt reply. I will try post-processing. By the way, is there any way to speed up and reduce memory? I am suffering from high time and memory consumption in a big dataset. Thank you very much.

We are working on a batch processing feature for seq2pat, in order to improve the speed and memory performance. We will have a new release once this capability gets ready and tested well. Meanwhile, you might also try to add stronger constraints to further reduce the search space or consider a smaller sampling of data, e.g. a sampling of sequences in your case, to speed up the mining on large data set.

from seq2pat.

Thanks. Looking forward to the new feature.

from seq2pat.