Genetic Algorithms (or optimization) in R

genetic-algorithm, optimization, r

Solution

Your problem is a Linear Programming (LP) task and should be solved as such. R has several packages for linear programming, here I will use `lpSolve`.

I put your data into a data frame `D` (299 workers).

str(D)
# 'data.frame': 299 obs. of  4 variables:
#  $ ID   : int  1 2 3 4 5 6 7 8 9 10 ...
#  $ Type : Factor w/ 4 levels "Cleanup Guy",..: 4 4 4 4 4 4 4 4 4 4 ...
#  $ Cost : int  5100 3900 5000 4500 6700 5000 4000 3500 7500 4500 ...
#  $ Value: int  25 18 23 20 29 21 17 15 32 19 ...

# Prepare constraint matrix
A <- matrix(0, nrow = 5, ncol = 299)
A[1, c(1:27, 279:299)] <- 1     # Plumbers
A[2, 28:97] <- 1                # Electricians
A[3, 98:190] <- 1               # Const Workers
A[4, 191:278] <- 1              # Cleanup
A[5, ] <- D$Cost                # cost <= 100000

# Prepare input for LP solver
objective.in <- D$Value
const.mat <- A
const.dir <- c(">=", ">=", ">=", ">=", "<=")
const.rhs <- c(1, 2, 2, 2, 100000)

# Now solve the problem
require(lpSolve)
sol <- lp(direction = "max", objective.in, # maximize objective function
        const.mat, const.dir, const.rhs,   # constraints
        all.bin = TRUE)                    # use binary variables only

# View the solution
sol
## Success: the objective function is 589
sum(D$Cost[inds])
## 100000
inds <- which(sol$solution == 1)
D[inds, ]
##      ID         Type Cost Value
## 1     1      Plumber 5100    25
## 51   51  Electrician 5000    24
## 54   54  Electrician 3500    16
## 101 101 Const Worker 3500    18
## 102 102 Const Worker 4400    21
## 103 103 Const Worker 5800    27
## 147 147 Const Worker 5600    29
## 149 149 Const Worker 3700    17
## 196 196  Cleanup Guy 6200    30
## 197 197  Cleanup Guy 4200    20
## 256 256  Cleanup Guy 4000    27
## 263 263  Cleanup Guy 3800    22
## 264 264  Cleanup Guy 3800    33
## 266 266  Cleanup Guy 3700    23
## 267 267  Cleanup Guy 4200    30
## 273 273  Cleanup Guy 3600    35
## 274 274  Cleanup Guy 5900    28
## 275 275  Cleanup Guy 3700    28
## 282 282      Plumber 7000    37
## 287 287      Plumber 3800    30
## 297 297      Plumber 3600    37
## 298 298      Plumber 5900    32

For explanation of the inputs see the help page of the `lp()` function.

Problem

I am trying to explore the area of optimization using GAs in R. I realize that R is probably not the best place for me to do this but that is what I know right now. So here is the problem: I have a set of people that I can pick to build my house most efficiently. I am willing to spend a maximum of 100,000 for the job (the limit beyond which the solution is no longer valid) and I HAVE to pick one electrician, two plumbers, two general construction workers, and two cleanup guys from a list of 500 or so (total 7 guys for the job). Each of these people have a rate they will charge for the job and overall value they will provide towards the job at hand (the sum total value needs to be maximized). See the data here: Data I have looked at R packages genalg and DeOptim where you can use the binary functions to select people that will maximize the overall value for a given price BUT I cant figure out how to put a limit on what type of worker is selected. The solution can be all electricians and while they all provide great value, they cant complete the job. So basically I am looking for a way to control the population and none of these packages seem to allow that. Does any have ideas on how i can solve this problem? I am sure many of you have looked at variations of this problem and have excellent insight. Thanks a lot in advance for your help. Here is the code I have obtained from forums here: ``` library(genalg) iter = 10 population = 500 # import csv file brank = read.csv("GA_Data.csv") dataset <- data.frame(item = brank$Person.ID, survivalpoints = brank$Value, weight = brank$Labor.Cost) weightlimit <- 100000 monitor <- function(obj) { minEval = min(obj$evaluations); plot(obj$mean, obj$best, type="p", main = obj$iter); } evalFunc <- function(x) { current_solution_survivalpoints <- x %*% dataset$survivalpoints current_solution_weight <- x %*% dataset$weight if (current_solution_weight > weightlimit) return(0) else return(-current_solution_survivalpoints) } GAmodel <- rbga.bin(size = length(brank$Person.ID), popSize = population, iters = iter, mutationChance = 0.01, monitorFunc = monitor, elitism = T, zeroToOneRatio = 200, evalFunc = evalFunc) cat(summary.rbga(GAmodel)) ######################## # STOP and replace "space" with "comma" in a text file solution = c(0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,1,0,0) # dataset[solution == 1, ] brank[solution == 1, ] sum(brank$Value[solution == 1]) sum(brank$Labor.Cost[solution == 1]) ```

Original source