How to properly declare a Function() for fitness in GA package?

user_14340450 0
Viewed 169

I started doing Linear Regression in RStudio but I'm just not able to figure out how to declare the fitness function required in the GA package that I pretend to use.

It shows the problem is related to the class object of the function, according to the following recurring error:

Error in (function (cl, name, valueClass) : assignment of an object of class “list” is not valid for @‘fitness’ in an object of class “ga”; is(value, "numericOrNA") is not TRUE

However, I've already tried changing the type of the train_set, with no success. I don't really know if the issue is more related to the proper management of two independent variables used.

Here is what I'm trying:

library(GA)

#from train, which is a list of 4 variables(total columns) and 80 lines
#Let's take the first 4 lines to exemplify

i  j k oil_cumulative_production
12 10 1                    2201.7
15 14 1                    2192.6
13  5 1                    2190.1
 9  4 1                    2185.3
15 11 1                    2193.1

train_set <- data.frame(y1 = train$oil_cumulative_production, x1 = train$i, x2 = train$j)
f <- function(x1, x2) { lmNp <- lm(y1 ~ x1 + x2, data = train_set)}
GA <- ga(type = "real-valued", fitness = function(x) f(x[1], x[2]), lower = c(1, 1), upper = c(20, 20), popSize = 50, maxiter = 100, optim = TRUE)

Any thoughts?

genetic-algorithm r optimization
1 Answers

If you want to maximize 'oil_cumulative_production', first train your regression model, then maximize by saying that the prediction made by the linear model using x1 and x2 is what you want to maximize.

Here's the code:

library(GA)
#> Loading required package: foreach
#> Loading required package: iterators
#> Package 'GA' version 3.2
#> Type 'citation("GA")' for citing this R package in publications.
#> 
#> Attaching package: 'GA'
#> The following object is masked from 'package:utils':
#> 
#>     de

train <- readr::read_table("i   j    k  oil_cumulative_production
12 10   1                    2201.7
15 14   1                    2192.6
13  5   1                    2190.1
9  4    1                    2185.3
15 11   1                    2193.1")

train_set <- data.frame(y1 = train$oil_cumulative_production, x1 = train$i, x2 = train$j)
lmNp <- lm(y1 ~ x1 + x2, data = train_set)
f <- function(x) {
    df4pred <- data.frame(x1 = x[1], x2 = x[2])
    predict(lmNp, df4pred)
}
GA <- ga(type = "real-valued", fitness = f, 
         lower = c(1, 1), upper = c(20, 20), popSize = 50, maxiter = 100, optim = TRUE)
plot(GA)

GA plot

GA@solution
#>      x1 x2
#> [1,]  1 20
GA@fitnessValue
#> [1] 2210.604
user_84854030
Related