Max SP algorithm doesn't seem to work in SPMF. It just gives me empty results no matter what I put to it.
Also, there's a bug where it demands me to create the new file for the results instead of creating it automatically like all other algorithms do.

VMSP won't open the created file automatically, btw.

For MaxSP, thanks for reporting the problem. It was not well integrated with the GUI. I have fixed that and will update the code on the website this week-end with a new release of SPMF.

For the other problem about VMSP, could you send me a sample input file and the parameters that you used, to my e-mail?

philippe.fv AT gmail.com

It will help me to fix the problem if there is one. On my computer, it currently seems to work. But performance that it depends on your data. Or perhaps that the bug has been fixed in my local version of SPMF.

Thanks



Edited 1 time(s). Last edit at 01/30/2015 01:34PM by webmasterphilfv.

Sorry I didn't answer earlier, was away from science for a while.
I've just gave my first tests to your new version of SPMF, and I'm very grateful that you've added maxgap parameter!
Both problems seem to be solved now. Hovever I've found something strange in another algo, VGEN.

Here's my data.
0 -1 0 -1 2 -1 4 -1 5 -1 7 -1 8 -1 9 -1 0 -1 11 -1 1 -1 6 -1 31 -1 32 -1 33 -1 26 -1 33 -1 26 -1 33 -1 26 -1 34 -1 35 -1 36 -1 37 -1 0 -1 26 -1 -2
0 -1 0 -1 2 -1 4 -1 5 -1 7 -1 8 -1 9 -1 0 -1 11 -1 1 -1 6 -1 25 -1 26 -1 25 -1 26 -1 31 -1 38 -1 32 -1 25 -1 26 -1 39 -1 34 -1 35 -1 36 -1 37 -1 0 -1 26 -1 -2
0 -1 1 -1 2 -1 3 -1 4 -1 5 -1 6 -1 0 -1 7 -1 8 -1 9 -1 10 -1 0 -1 11 -1 1 -1 6 -1 0 -1 14 -1 40 -1 41 -1 42 -1 14 -1 0 -1 32 -1 -2

Params:
support 1
maxgap 3

Result on VGEN:
SUP 3
(no patterns in the output, but the support is there)

I've also noticed that the statistics for this experiment state: "minsup 3", it probably meant "maxgap 3"
Tried it with significantly bigger dataset, the program freeze.

Actually, this is normal.

If you see a line such as :

SUP 3

it means that it is the empty set. The empty set has a support of three and is a generator so it should be output.

In general, in sequential pattern mining, we are not interested by the empty set, but when we talk about generators, we should not forget the empty set because it is often a generator. For some applications related to generators, the empty set is important So that is the reason why you will see the empty set in the results of VGEN when the empty set is a generator.

Best,



Edited 2 time(s). Last edit at 02/08/2017 06:39PM by webmasterphilfv.

Thank you for your answer.

In the conclusion, can I ask you if the tendency of the latest update will continue? I mean, do you plan to add restrictive options to other seq.pattern mining algos any soon? What interests me most is closed pattern mining. On this answer, my current research strategy depends, you see. Like, if it's not coming soon, then I'll make some post-processing code for normal seq.pattern mining results, or something.

Hi Aidir,

You can tell me which features are the most important for you and for which algorithm and I will try to add them. Some features are easier to be added, and it is easier to add features on some types of algorithms than for some others, so I need to also consider that.

Until next monday, I will not be able to work on SPMF. But next week would be a good time for me to add new features.

Best,

Hello,
I didn't really hope for such luck

My top priority is maxgap parameter - added to any closed pattern mining algorithm, which is effective on memory. I'm not sure which one is the least memory-consuming, since the algorithms are usually tested in terms of speed.
It seems that Bide+ isn't very memory consuming one, and if that's right, then I'd name this one.
If applying maxgap is somehow not possible with closed pattern mining algorithms, I'd appreciate having an option to setmax pattern length.