Take the 2-minute tour ×
Mathematica Stack Exchange is a question and answer site for users of Mathematica. It's 100% free, no registration required.

I know how to model this question on paper, but I don't know how doing it in Mathematica.

Imagine , there is a university student who want to choose courses for this semester and he got the university program chart that looks like below in excel format :

University program chart

Lecturers Rating is out of 10 and it means which lecturer is giving a better score in the final exam and teaches better.

Course Credit is a restriction, each semester each student should choose courses that total of course credit must be greater than 7 and less than 12:

It is not allowed to pick two courses with the same name.

He/She wants to optimize two things :

  1. He/She wants to maximize course credits that he/she could use (12)
  2. He/She wants to pick up courses with better Lecturers

Also He/She wants to get the result like a schedule table like this:

Here is excel File form university.

share|improve this question
    
The code that generates the schedule you've posted a picture of is available here. –  Pickett Jan 19 at 23:01
    
@Pickett Yes, I'll edit the question. And I know where is the, But I want to do it automate. –  DSaad Jan 19 at 23:04
    
how much larger? The problem size grows very fast with the number of courses. With 11 you can redily chack all the combintaions, but with even 100 you'd need an entirely different approach. –  george2079 Jan 21 at 23:37
    
@george2079 maximum courses are 15 and maximum lecturers are 4 for each course. –  DSaad Jan 22 at 0:33

2 Answers 2

up vote 9 down vote accepted
+50

Here is an ILP approach. It can be modified to alter requirements e.g. if a course has a lab, must take neither or both, maybe insist on at most one instructor with the lowest rating, at most two classes before 9 AM, have courses that meet on multiple days, etc.

I entered it all by hand although clearly one could use Import and further processing.

courses = {{"math", 3, "M", {8, 10}, 5}, {"de", 3, "Th", {8, 10}, 
    8}, {"chem", 2, "M", {8, 10}, 9}, {"physL", 1, "Th", {9, 10}, 
    4}, {"de", 3, "F", {13.25, 15.25}, 6}, {"chem", 2, 
    "F", {13.25, 15.25}, 9}, {"chemL", 1, "W", {9, 10}, 10}, {"physL",
     1, "W", {9, 10}, 7}, {"phys", 3, "M", {10.25, 12.25}, 
    6}, {"phys", 3, "W", {10.25, 12.25}, 5}, {"math", 3, 
    "Th", {10.25, 12.25}, 7}};

vars = Array[v, Length[courses]];
obj = vars.courses[[All, -1]];
c1 = Map[0 <= # <= 1 &, vars];
c2 = {Element[vars, Integers], 7 <= vars.courses[[All, 2]] <= 12};
c3 = Flatten[
    Table[If[
      courses[[j, 3]] == courses[[k, 3]] && 
       IntervalIntersection[Interval[courses[[j, 4]]], 
         Interval[courses[[k, 4]]] /. 
          Interval[{aa_, aa_}] :> Interval[]] =!= Interval[], 
      vars[[j]] + vars[[k]] <= 1]stronger
     , {j, 1, Length[vars] - 1}, {k, j + 1, Length[vars]}]] /. 
   Null :> Sequence[];
c4 = Flatten[
    Table[If[courses[[j, 1]] == courses[[k, 1]], 
      vars[[j]] + vars[[k]] <= 1]
     , {j, 1, Length[vars] - 1}, {k, j + 1, Length[vars]}]] /. 
   Null :> Sequence[];
constraints = Union[Join[c1, c2, c3, c4]];

With this set up we can use FindMaximum and the like.

{max, sched} = FindMaximum[{obj, constraints}, vars];

max

(* Out[259]= 40. *)

vars /. sched

*(* Out[260]= {0, 1, 0, 0, 0, 1, 1, 0, 1, 0, 1} *)

Pick[courses, vars /. sched, 1]

(* Out[262]= {{"de", 3, "Th", {8, 10}, 8}, {"chem", 2, 
  "F", {13.25, 15.25}, 9}, {"chemL", 1, "W", {9, 10}, 10}, {"phys", 3,
   "M", {10.25, 12.25}, 6}, {"math", 3, "Th", {10.25, 12.25}, 7}} *)

( Same but with Maximize)

Maximize[{obj, constraints}, vars]

(* Out[273]= {40, {v[1] -> 0, v[2] -> 1, v[3] -> 1, v[4] -> 0, v[5] -> 0,
   v[6] -> 0, v[7] -> 1, v[8] -> 0, v[9] -> 1, v[10] -> 0, 
  v[11] -> 1}} *)

To find all schedules that are tied on the objective function one could use Reduce.

Reduce[Flatten[{obj == 40, constraints}], vars]

(* Out[275]= (v[1] == 0 && v[2] == 1 && v[3] == 0 && v[4] == 0 && 
   v[5] == 0 && v[6] == 1 && v[7] == 1 && v[8] == 0 && v[9] == 1 && 
   v[10] == 0 && v[11] == 1) || (v[1] == 0 && v[2] == 1 && v[3] == 1 &&
    v[4] == 0 && v[5] == 0 && v[6] == 0 && v[7] == 1 && v[8] == 0 && 
   v[9] == 1 && v[10] == 0 && v[11] == 1) *)

Since this is all ILP under the hood I would not expect it to handle huge problems. Offhand I do not have a good guess as to how far it might scale.

Another thing to note is that I made no effort to get the maximum advantage from avoiding conflicts. I only looked at pairs of classes. Triples that have nontrivial meeting time intersection would give rise to tighter (that is, more restrictive) inequalities of the form x+y+z<=1. Such better inequalities could make a difference in how far one might scale this approach.

share|improve this answer
    
Could you upload your file, because when I run your codes I get a lot of errors? –  DSaad Jan 25 at 0:44

It seems that maximization of ratings leads to maximization of credits using. My code (but not answer) below. It's sequential search but it works.

data = Import["university program chart.xlsx"];
dtt[d_] := 
  d /. {"Monday" -> 0, "Tuesday" -> 1 24 60 60, 
    "Wednesday" -> 2 24 60 60, "Thursday" -> 3 24 60 60, 
    "Friday" -> 4 24 60 60, "Saturday" -> 5 24 60 60};
strpr[line_] := Module[{ret = {0, 0, 0, 0, 0}},
   ret[[1]] = 
    Interval[(AbsoluteTime[{#, {"Hour", ":", "Minute"}}] & /@ 
        StringSplit[line[[5]], "-"]) + dtt[line[[4]]]];
   ret[[2]] = Round[line[[1]]];
   ret[[3]] = Round[line[[3]]];
   ret[[4]] = Round[line[[7]]];
   ret[[5]] = line[[2]];
   Return[ret];
   ];
ndata = strpr /@ (data[[1, 2 ;;]]);
mdata = Subsets[ndata][[2 ;;]];
kdata = Select[mdata, 
   7 <= Total[((#)\[Transpose])[[3]]] <= 12 && 
     Not[Or @@ 
       IntervalMemberQ @@@ 
        Subsets[Flatten[(#)\[Transpose][[1]]], {2}]] && 
     DuplicateFreeQ[#\[Transpose][[5]]] &];
util = Total[(#\[Transpose])[[4]]] & /@ kdata;
listofnumbers = 
  kdata[[#]]\[Transpose][[2]] & /@ Flatten[Position[util, Max[util]]];

And the result is:

TableForm /@ (data[[1, # + 1]] & /@ listofnumbers)

Output:

2 Differential Equations 3. Thursday 8:00-10:00 Dr. Smith 8.

3 Chemistry 2. Monday 8:00-10:00 Dr. Cho 9.

7 Chemistry Lab 1. Wednesday 9:00-10:00 Dr. Xaviers 10.

9 Physics 3. Monday 10:15-12:15 Dr. Rosta 6.

11 Math 3. Thursday 10:15-12:15 Dr. Jones 7.

Or

2 Differential Equations 3. Thursday 8:00-10:00 Dr. Smith 8.

6 Chemistry 2. Friday 13:15-15:15 Dr. Xaviers 9.

7 Chemistry Lab 1. Wednesday 9:00-10:00 Dr. Xaviers 10.

9 Physics 3. Monday 10:15-12:15 Dr. Rosta 6.

11 Math 3. Thursday 10:15-12:15 Dr. Jones 7.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.