Saving Files With Slightly Different Names-CodePudding

I have the following loop that runs 5 different Random Forest Models on 5 different datasets (these datasets are in a list object):

# Base Loop : Works Fine
results_1 <- list()
results_2 <- list()

for (i in 1:5){
    model_i <- randomForest(class ~  height   weight   salary, data = X[[i]])

    predict_i <- data.frame(predict(model_i , test_set ,type="prob"))
    predict_i$id = 1:nrow(predict_i)
    
    results_1[[i]] <- model_i
    results_2[[i]] <- predict_i 
}

I would like to save each of these models (in "my documents") using the "model_i.RDS" format. If this were to work, there would be 5 RDS files (model_1.RDS, model_2.RDS, model_3.RDS, model_4.RDS, model_5.RDS) in saved/created in "my documents".

I thought R would pick up on this by itself seeing as I have defined the "index i" already:

# Method 1: Does Not Work
results_1 <- list()
results_2 <- list()

for (i in 1:5){
    
    
    model_i <- randomForest(class ~  height   weight   salary, data = X[[i]])

    saveRDS(model_i, "model_i.RDS") 
    
    predict_i <- data.frame(predict(model_i , test_set ,type="prob"))
    predict_i$id = 1:nrow(predict_i)
    
    results_1[[i]] <- model_i
    results_2[[i]] <- predict_i
}

But this is only saving a single "RDS" file.

I then tried to be more explicit with the saving command:

# Method 2: Also Not Working
wd = getwd()

results_1 <- list()
results_2 <- list()

for (i in 1:5){
    model_i <- randomForest(class ~  height   weight   salary, data = X[[i]])
    
    saveRDS(model_i, paste0("wd", paste("model_", i, ".RDS")))
            
    predict_i <- data.frame(predict(model_i , test_set ,type="prob"))
    
    predict_i$id = 1:nrow(predict_i)
    
    results_1[[i]] <- model_i
    results_2[[i]] <- predict_i
}

But this still isn't working (a single file is being saved instead of 5 files).

Can someone please show me how to fix this problem?

Note : Sample Data for Problem:

 library(randomForest)
    
    test_set = data.frame( class = as.factor(sample(c(0,1), replace=TRUE, size=100)), height = rnorm(100,100,100), weight = rnorm(100,100,100), salary = rnorm(100,100,100)) 
    
    train_data_1 = data.frame( class = as.factor(sample(c(0,1), replace=TRUE, size=100)), height = rnorm(100,100,100), weight = rnorm(100,100,100), salary = rnorm(100,100,100)) 
    
    train_data_2 = data.frame( class = as.factor(sample(c(0,1), replace=TRUE, size=100)), height = rnorm(100,100,100), weight = rnorm(100,100,100), salary = rnorm(100,100,100)) 
    
    train_data_3 = data.frame( class = as.factor(sample(c(0,1), replace=TRUE, size=100)), height = rnorm(100,100,100), weight = rnorm(100,100,100), salary = rnorm(100,100,100)) 
    
    train_data_4 = data.frame( class = as.factor(sample(c(0,1), replace=TRUE, size=100)), height = rnorm(100,100,100), weight = rnorm(100,100,100), salary = rnorm(100,100,100)) 
    
    train_data_5 = data.frame( class = as.factor(sample(c(0,1), replace=TRUE, size=100)), height = rnorm(100,100,100), weight = rnorm(100,100,100), salary = rnorm(100,100,100)) 
    
# data used in question
 X = list(train_data_1, train_data_2, train_data_3, train_data_4, train_data_5)

CodePudding user response：

In the last attempt, the OP used paste, but there was no / between the working directory and file name - safer to use file.path. In addition, the wd used was "wd" which will be taken literally instead of the value stored in the object. Instead, it can be paste0(wd, "/model_", i, ".RDS")

results_1 <- vector('list', 5)
results_2 <- vector('list', 5)
for (i in 1:5){
    model_i <- randomForest(class ~  height   weight   salary, data = X[[i]])
    
    saveRDS(model_i, file.path(wd, paste0("model_", i, ".RDS")))
            
    predict_i <- data.frame(predict(model_i , test_set ,type="prob"))
    
    predict_i$id = 1:nrow(predict_i)
    
    results_1[[i]] <- model_i
    results_2[[i]] <- predict_i
}

-output