Mega request

[ellebi]

I chose this subject because I'm aware that what I'm going to propose is a really huge feature.

In the timetables I have to elaborate, I often have problems related to shortage of first or last hours, expecially on Friday or Monday (caused by teacher's constraints), but I don't have a tool that say me precisely which days/hours are the most critical, so I "expand" constraints quite blindly....
If I see at the most critical activities during the generation I have suggestions on which are the critical student's groups, but they aren't so precise regarding the slot with few or null teacher's availabilities...

I thought a lot on a tools that could help to find why a timetables it's impossible or near impossible: so, in one sentence I'm wondering if it would be possible to calculate the probability that a slot (day+hour: "Monday 8:00-9:00") will be destinated to any student's group?

If I see this problem from far, I would say yes, we "only" need to calculate the probability with which each teacher will occupy this slot (basing on their constraints), combined with all other constraints concerning the Student's group, Subjects, etc.

The algorithm should consider each constraint inserted in dataset, calculate the probability for a slot to be assigned to someone and store it. After that, it will combine all calculated probabilities so to generate a big probability table...

I imagine a multicolor table, with time in horizontal and student's groups in vertical, with light or strong "red" and "green slots based on this probability

There are many cases to examine, but I don't think it's impossible to do, it's "only" (really) time espensive to develop....

In addition, I think it would be possible to "pivot" the data, so to see the probability by teacher (always to occupy that slot), or by other dimensions...

I don't have the skills to do that (and firstly to estime its feasibility), but maybe someone has the possibility to start the project...

It would be possible to test it starting with a first set of constraints considered in calculations, and after enlarge to other constraints typology....

Any ideas / comments are welcome!

I swear, I'm not crazy, it's a serious suggestion :-D

[Volker Dirr]

Sadly a very difficult task. We already now some very "easy" datasets. (So very low number of activities and very low number of constraints.) Even in that samplefiles it is very difficult to write an algorithm that can see the problem, even it is very easy to see the problem as a human.

[Liviu Lalescu]

Yes, like Volker said, to check that a timetable is possible you need to run FET a long time on that data 

Maybe your ideas can somehow be put into practice and work in some cases. I read your message more times, but I am still not sure it is always possible. Maybe something is sometimes possible, considering the interaction of all the constraints.

But there is also the financial part. I am grateful that you donated so kindly, but the donations/payments are very few. I still hope that a larger entity interested into FET could consider this. I can give more details about this, if anybody is interested.

[ellebi]

Dear Liviu and Volker,
thank you for your replies.
I totally understand what you said, and I agree: it's a "big stuff"
My primary intention was to create interest in this, and I hope someone skilled can offer his help.
By my part, if I can do something, I'll do (I'm not a real programmer, but I have some little skills), feel free to propose...

In the meantime, when I'll have time to do that, I'll send you some details about my idea, based on statistical calculations and not "observations" of many runs. I repeat, I can only describe behaviours and not write real algorithm, because I haven't all the skills needed, but (who knows) it can be a first little step

I agree also on what you said about financial part, and I hope it can go better in the future, you deserve it (the rows in donation page are really too few ).

By the way, if you have some tool to help me in finding the most critical slots, you are welcome!

Many thanks, really, for your work.

Bye!

[Liviu Lalescu]

Thank you, Luca,

Unfortunately, we have no such tool. If you send more details about your idea, I might try something, but I am not sure it will come out useful in general.

[ellebi]

Ok, Liviu
When I have the time to do it, I'll prepare some details about my idea and I'll send it to you.
Have a nice day.
Luca

[Liviu Lalescu]

Thank you, Luca, you too!