Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Employee Scheduling

developers.google.com

Post reply on HN
Print view View original

101–110 of 137 posts

Re: Employee Scheduling

#101

Post by mochomocha »

jonbaer wrote: post #100

Maybe a little OT, but why aren't the OR-tools in general more focused (or more "marketed") towards GPU/TPU clouds since they seem like the exact problems which they are made for? More focus seems to be towards TF/PT/NN and not the real problems like what OR-tools and solvers in general push for.

Typical algorithms solving combinatorial problems are usually heavily branching, so not so much vectorizable. For example for a branch-and-cut MIP solver, you could run the LP-relaxed problems on GPU, but most of the logic lives in branching and cutting heuristics that are much harder to parallelize.

Re: Employee Scheduling

#102

Post by 7thaccount »

jonbaer wrote: post #100

Maybe a little OT, but why aren't the OR-tools in general more focused (or more "marketed") towards GPU/TPU clouds since they seem like the exact problems which they are made for? More focus seems to be towards TF/PT/NN and not the real problems like what OR-tools and solvers in general push for.

A lot of OR stuff can be parallelized, but it isn't a panacea. Those techniques don't necessarily get you the best answer sometimes. My company does some massive optimization problems and although we have lots of cores for all the simulations, any one is pretty much single core + a lot of RAM. It doesn't work well on GPU like some heuristic optimization problems can.

Edit: Someone smarter than me says it better below :)

Re: Employee Scheduling

#103

Post by dwaltrip »

marcosdumay wrote: post #71
duxup wrote: post #66

Earlier quoted context omitted.

Humans might be kinda slow and expensive, but their ability to handle a lot of variable data and inputs and weigh them dynamically, filter out absurd results before applying them, and etc is pretty amazing. Like if someone comes to me and says "what if we... did X with the schedule". I could tell them how to do that and not to right away. The code would have to be changed / rewritten and tested and fail at it each ti…

> their ability to handle a lot of variable data and inputs and weigh them dynamically I dunno. Computers are pretty good at this one too, on different kinds of variance and weights. The largest problem I see is that everyone is discussing "either or" on this thread, while the clear winner is both.

They are far worse at it than humans. This is the difference between narrow AI and general AI. Humans can do millions of different complex tasks, and can adapt when the situation changes. Computers are terrible at updating their models automatically based on new information, as they don't have understanding of the bigger context. Worse still, software today has no way of analyzing and adjusting the assumptions it is operating off of.

Re: Employee Scheduling

#104

Post by matchagaucho »

cosmodisk wrote: post #60
duxup wrote: post #54

Technical solution aside: I wrote a schedule for a team of ~20 co-workers for a 24/7 support shop. This required people change shifts / sleep schedules often. The schedule was a point of contention for years. After I volunteered to write the schedules, in 3 months I had everyone happy. It was not hard.....but.... The catch was that I did it manually for a month at a time with a wide ranging amount of changing priorit…

One of the senior teachers in my school used to organise schedule for 1000 pupils and 50+ teachers. With some minor things everything worked pretty damn well.Then they decided to get some software to do the same.Never ended up working...

It may also be the software started enforcing things like limited overtime and back-to-back shifts that humans were otherwise fine scheduling.

Re: Employee Scheduling

#105

Post by cosmodisk »

matchagaucho wrote: post #104
cosmodisk wrote: post #60

Earlier quoted context omitted.

One of the senior teachers in my school used to organise schedule for 1000 pupils and 50+ teachers. With some minor things everything worked pretty damn well.Then they decided to get some software to do the same.Never ended up working...

It may also be the software started enforcing things like limited overtime and back-to-back shifts that humans were otherwise fine scheduling.

As a problem, classroom scheduling is very interesting. The kids and the teachers were well aware of challenges anyone doing the scheduling hadv and very much doubt any software at the time could handle non standard situations. Interestingly enough,just last year I was trying to find off the shelve software that could do scheduling and have a calendar feature for one our suppliers. After considering 10 or so options I ended up with a conclusion that unless someone would do custom solution, they'd have to continue using Excel for it.

Re: Employee Scheduling

#106

Post by cosmodisk »

watwut wrote: post #99
duxup wrote: post #66

Earlier quoted context omitted.

Humans might be kinda slow and expensive, but their ability to handle a lot of variable data and inputs and weigh them dynamically, filter out absurd results before applying them, and etc is pretty amazing. Like if someone comes to me and says "what if we... did X with the schedule". I could tell them how to do that and not to right away. The code would have to be changed / rewritten and tested and fail at it each ti…

I suspect that the big thing there is negotiation and social skills. It is not just about making schedule, it is also about talking with people, making them compromise, making them happy despite not gaining everything they wanted, explaining them that what they wanted they wont like actually, making them understand others too.

This is a good point on social aspect: no software can go to a couple of teachers and say: look, there's an overlap and we've only got one classroom,how about you do x,y,z and then next time you get x,y,z.

Re: Employee Scheduling

#107

Post by nitwit005 »

duxup wrote: post #66
cosmodisk wrote: post #60

Earlier quoted context omitted.

One of the senior teachers in my school used to organise schedule for 1000 pupils and 50+ teachers. With some minor things everything worked pretty damn well.Then they decided to get some software to do the same.Never ended up working...

Humans might be kinda slow and expensive, but their ability to handle a lot of variable data and inputs and weigh them dynamically, filter out absurd results before applying them, and etc is pretty amazing. Like if someone comes to me and says "what if we... did X with the schedule". I could tell them how to do that and not to right away. The code would have to be changed / rewritten and tested and fail at it each ti…

Assuming you can get someone highly competent to do scheduling, which isn't all that likely. More likely you'll get someone who both produces mediocre results, and favors their friends.

Even if you can get competence, you may want to automate it anyways. I'm currently working in health, and I recently visited a radiology center that claimed it took a full year to get front desk staff fully up to speed. Understanding the full scheduling requirements for all scans, some medicare rules that influence scheduling, and how to handle all the various problems, takes a long time. They live in terror of having front desk staff leave, because a quick replacement is impossible.

Re: Employee Scheduling

#108

Post by icebraining »

cosmodisk wrote: post #106
watwut wrote: post #99

Earlier quoted context omitted.

I suspect that the big thing there is negotiation and social skills. It is not just about making schedule, it is also about talking with people, making them compromise, making them happy despite not gaining everything they wanted, explaining them that what they wanted they wont like actually, making them understand others too.

This is a good point on social aspect: no software can go to a couple of teachers and say: look, there's an overlap and we've only got one classroom,how about you do x,y,z and then next time you get x,y,z.

Well, not yet.

Re: Employee Scheduling

#109

Post by duxup »

freepor wrote: post #88
duxup wrote: post #54

Technical solution aside: I wrote a schedule for a team of ~20 co-workers for a 24/7 support shop. This required people change shifts / sleep schedules often. The schedule was a point of contention for years. After I volunteered to write the schedules, in 3 months I had everyone happy. It was not hard.....but.... The catch was that I did it manually for a month at a time with a wide ranging amount of changing priorit…

Fascinating. Can you give some examples of the priorities you incorporated that would be hard for off-the-shelf software to digest?

The biggest challenge is just addressing the constantly shifting math on what is 'good'.

So late night shifts are generally undesirable, but required to be covered and so you really want to spread them out. Having said that for some folks they'd rather do a bunch in a row, others spread them out. Some folks mind them less than others.

Just making the math decide that an even number across the whole team is 'fair' always leads to some weirdness no matter how it works out.

So a weekend night shift might be 3 days of work then 4 off. That's awesome for a guy whose vacation starts the following week because he gets more time off. He might actually WANT that. So in that case a night shift is desirable and makes the guy happier.

Another person might not want that or not have a vacation... so his point of view might differ. But maybe next weekend he feels differently.

And that's just one given weekend shift with somewhat unpredictable results.

The real key though is not just noting how the variables change wildly it is that keeping those guys happy, being flexible for them... makes them MORE willing to take a rough shift now and then. That's even HARDER to quantify, but EXTREMELY valuable.

And of course my ability to walk up to a dude, explain the situation, he trusts me because I've done this before / him favors is just something a computer that fires off an email just can't do.

Re: Employee Scheduling

#110

Post by mighty_plant »

Google or-tools are quite fun although a little bit of a pain to setup and get started at least if you want to use the JVM and support different OS. Just to share a little anecdote about my use case: The department I work in consists of about 40 developers and once a year we had a day together just to form new teams in a self organized manner. Management wanted about one third of a team to move to a different team to spread and gather knowledge but completely backed off during that day. We had a special coach and scrum masters supporting us but still it was very exhausting and a lot of the developers didn't like it. The devs that showed motivation to change teams were more or less forced to go to the teams which were disliked the most (either because of their domain or the already present members). So in the end there was not much rotation and the whole thing was quite expensive so that this year it was cancelled. That's when I started to use Google or-Tools to write a little tool that suggests teams based on some constraints like size, at least one experienced dev / software architect per team and an optimization function based on preferences of each dev for a certain domain as well as their sympathy towards other devs When I announced it it got almost no attention and was dismissed as "inhuman".
Post reply on HN