What have you used OR for other than your "job"?

 5 1 I'm looking for answers that is using OR in your daily lives that is outside of your day job. Have you used OR for fun? Did you use it to solve for something around the house, community, or charity? Maybe you just have fun developing solutions to some sort of decision analysis. For me I'm a Fantasy Football (that is American Football) nut. I've developed some pretty basic forecasting models to predict player performances. I've also used OR to develop a method to assign players to a little league soccer team that I coach. My objective is equitable field time given limited number of players on the field at a given time. asked 27 Apr '10, 20:53 larrydag 1 ♦ 3.2k●5●13●26 accept rate: 9%

 5 My wife is a nurse and she worked for a service with only three nurses that made their shift schedule by hand every month. She asked me to develop a tool for them that maximize the attractiveness of the schedule (according to some preferences of them for certain shifts) and meets the hard constraints of total weekly and monthly time and service needs. I developed a small MIP to find their shifts. They were very happy using the new optimized shifts answered 28 Apr '10, 04:17 Juan G. Vill... 51●1 accept rate: 0% 1 Your model has a large market :) (28 Apr '10, 06:26) Mark ♦
 5 I'm not sure that this counts since I'm not a professional in OR, but when learning OR and constraint programming I have modeled a lot of small - and not so small - puzzles and recreational mathematics problems. Most of them are in MiniZinc (see http://www.hakank.org/minizinc/ ) and also Comet ( http://www.hakank.org/comet/ ), and some other CP systems. Some examples:  - almost all the IP puzzles by Martin Chlond - some Enigma problems - Sudoku, Hidato, Kakuro, KenKen, Strimko, and other grid puzzles - Minesweeper - Nonogram - etc  (I have a category for puzzles/recreational mathematics on my Constraint Programming blog: http://www.hakank.org/constraint_programming_blog/puzzles_recreational_mathematics/ ) answered 28 Apr '10, 05:03 Hakan Kjelle... 491●1●2●7 accept rate: 0%
 4 Continuing the sports angle, I provide the schedules for my son's soccer league (6 age groups, about 60 teams total, over 8 weeks). None of my paying clients have ever been so appreciative as this pro bono work! answered 27 Apr '10, 21:55 Michael Trick ♦♦ 4.1k●4●15●33 accept rate: 20% I've definitely thought of volunteering for my son's league. (28 Apr '10, 11:53) larrydag 1 ♦
 4 When I first moved to Pittsburgh I modeled my bike commute from home to university as a network flows model which minimized the amount of uphill / downhill travel needed on my bike (for those who don't know, Pittsburgh is rather hilly), using Google Earth to find the elevations for each potential intersection I could traverse. Heading in to work worked just fine (net decrease in elevation); turned out the bike/me combination didn't handle the reverse route well, I ended up walking my bike up one rather large hill, and that was the demise of my bike commute to work :) Maybe I'll get a better bike someday... answered 28 Apr '10, 01:28 Andy 41●1●1 accept rate: 0% 1 ... or a better model? :-) Maybe constrain maximum increase in elevation over any short segment? (28 Apr '10, 12:26) Paul Rubin ♦♦ Any success using Google Maps? I know it shows bike trails. I wonder if you could use the Google Maps API to develop optimal bike paths. hmmmmmm... (29 Apr '10, 18:22) larrydag 1 ♦
 3 I modelled sudoku as an IP once and used a solver to find the answer every morning so I could fill in the paper at work before my workmate (who loved sudoku) could get to it. I've also had a fair amount of experience developing predictive models for both sports competitions and horse racing. Not with any remarkable success yet, although I have hope for the horse racing models. I'm currently attempting to use metaheuristics to design trading algorithms for currency trading, and am cautiously optimistic about the potential. answered 27 Apr '10, 23:37 DC Woods ♦ 4.1k●2●25●46 accept rate: 5% I was interested in solving sudoku by LP, too :-) (28 Apr '10, 07:47) Karsten W.
 2 Funny you should mention the soccer team problem -- I'm working on an open-source application (ParityBuilder) to assign players to teams (at the league level) so as to achieve competitive balance. A rec league volunteer first approached me for some help; he's a chemist by training, so has some understanding of optimization models, and he was struggling to get it done in Excel. (Turns out Solver's variable limit is too small.) A few months ago, I helped out a buddy who was looking to schedule couples in a local duplicate bridge club. They meet once a month at multiple locations, rotating both who's playing whom and who is hosting each group, and they needed a balanced schedule. And once upon a time there was a computer game, written in Basic, called "Lunar Lander". You used the arrow keys to fire the retro rockets on a LEM, and the idea was to land gently before you ran out of fuel (or achieved escape velocity, never to be seen again). The game was in source code, so the physical model was visible to anyone curious enough to look. I used an optimal control model to minimize fuel consumption with zero velocity at touchdown, discretized it with respect to time, and solved the game. Of course, that ruined any future play value. answered 27 Apr '10, 21:46 Paul Rubin ♦♦ 14.6k●4●12 accept rate: 19% very interesting applications (27 Apr '10, 23:12) Mark ♦ You are able to upgrade the Excel solver, so that it can handle very large problems, using the same easy interface. I haven't used it but it looks quite good and relatively cheap: http://www.solver.com/ (27 Apr '10, 23:39) DC Woods ♦ I had a chance to play with the commercial version (or at least a commercial version) of Solver once. The catch was that the rec league folks have a budget of USD 0 for software. Also, in setting up the model there's a bunch of preprocessing that probably could be done with Excel macros, but not easily (and not by macro-averse me). (28 Apr '10, 12:29) Paul Rubin ♦♦
 2 I once wrote a little optimization script to play one of those flash games for me. The interface was a little buggy but at the end I was really proud of it. answered 27 Apr '10, 22:43 Mark ♦ 3.6k●2●23●50 accept rate: 9%
 2 I swim with the local Masters club and when it comes to swim meets picking relay teams is a bit tricky. There are many constraints on the make-up of the relay teams and things get even trickier when it comes to medley relay teams, where each of the four different strokes are swum by a different swimmer. I developed an IP that would make the best assignments. Of course, as with much else, in practice things were a little different: Masters swimmers are a fairly laid-back lot and sometimes the swimmers don't show up, in spite of registering, etc. The solutions that the IP gave tended to not be very robust and did not respond well to last-minute rejigging. For example, the generated solutions often found a team of swimmers that were just inside the allowable age limits. Finding a replacement for one of them tended to be more difficult than the coach's solution, which would usually have more slack. I could run it on the day, I suppose... That would allow me use more accurate input data, too! :-) answered 23 May '12, 04:43 Pa H 51●4 accept rate: 0%
 0 I created a GPML model for scheduling people at a reference desk for my wife's work. It worked well and I was rather proud of myself (I'm not an OR professional). Four issues, though: Complex to use. I created an Excel spreadsheet for a front end that would allow entering the schedules, requirements, etc, and that does sanity checks on the data. Then you export to CSV, then run glpsol which outputs a CSV which you import. A little complex for my wife's taste. Only one solution. It only outputs a single solution, while my wife would prefer to see a couple to choose from. (SOme of this is due to item #3's concern.) It doesn't take into account all of the things my wife does when she schedules by hand. For example, she avoids more than four consecutive hours, but also avoids more than two separated runs in a day, and there are people who like mornings and people who like afternoons, etc, etc. Someone takes vacation at the last minute. (My wife would prohibit this, but...) So the schedule needs to be rerun, but not from scratch: whatever part of the week's schedule that's already done needs to stay fixed and only the rest of the week should be variable. I can make some progress on 3, and I know what has to be done for 4, but... It's been quite fun and it works well. I'm looking for a larger problem that would run longer, so I can see if software I've compiled for multiple processors actually makes a difference. answered 30 Jun '12, 08:37 Wayne 1●1 accept rate: 0%
 toggle preview community wiki

By Email:

Markdown Basics

• *italic* or _italic_
• **bold** or __bold__
• image?![alt text](/path/img.jpg "Title")
• numbered list: 1. Foo 2. Bar
• to add a line break simply add two spaces to where you would like the new line to be.
• basic HTML tags are also supported

Tags:

×127
×4
×1