rfurrr

r future_map with ggpredict


I have generated 10 model summaries after fitting the same model on 10 different subsets of the dataset which is as follows

library(mice)
data("nhanes")
head(nhanes)

imp <- mice(nhanes, print = FALSE, m = 10, seed = 24415)
df <- complete(imp, "long")
model_fit <- lapply(1:10, function(i) {
  model = lm(bmi ~ lspline(age, c(5, 10)) + hyp + chl,
             data = subset(df, `.imp`==i))
})

From this I get different ggpredict objects

ggpredict(model_fit[[1]], c("age[all]", "hyp"))
ggpredict(model_fit[[2]], c("age[all]", "hyp"))
ggpredict(model_fit[[3]], c("age[all]", "hyp"))
ggpredict(model_fit[[4]], c("age[all]", "hyp"))
ggpredict(model_fit[[5]], c("age[all]", "hyp"))
ggpredict(model_fit[[6]], c("age[all]", "hyp"))
ggpredict(model_fit[[7]], c("age[all]", "hyp"))
ggpredict(model_fit[[8]], c("age[all]", "hyp"))
ggpredict(model_fit[[9]], c("age[all]", "hyp"))
ggpredict(model_fit[[10]], c("age[all]", "hyp"))

I am looking for an efficient way to a) Estimate the average of all the ggpredict objects by hp and age

I am trying to use future_map function with ggpredict. This is my code

library(lspline)
library(purrr)
library(furrr)

plan(multisession, workers =20)
test1 <- future_map(model_fit, ggpredict,  c("age[all]", "hyp"), library(lspline), progress = TRUE)

I get a warning and an error message

Warning: could not access model information. Warning could not find any variables for the specified model component.

i In index = 1.

i With name: modelStruct

Caused by error in !is.null(model_Info) && model_info$is_trial: ! invalid y type in x && y

However if I switch to regular map function, there is no error or warning everything works fine, it just takes too long.

test1 <- map(model_list[[1]] , ggpredict,  c("age[all]", "hyp"))

I like to know where I am going wrong or the right way to specify future_map to work with ggpredict. Thanks in advance.


Solution

  • library(lspline) inside future_map() may be causing the problem.

    The following works:

    (Sometimes it is necessary to explicitly load a package for each worker using packages = c("package_name") as an extra argument.)

    [Updated for the second model with lspline.]

    library(mice)
    library(ggeffects)
    library(tidyverse)
    library(furrr)
    
    imp <- mice(nhanes, print = FALSE, m = 10, seed = 24415)
    df <- complete(imp, "long")
    model_fit <- lapply(1:10, function(i) {
      model = lm(bmi ~ age + hyp + chl,
                 data = subset(df, `.imp`==i))
    })
    
    plan(multisession, workers = 10)
    
    future_map(model_fit, ggpredict, c("age[all]", "hyp"), progress = TRUE) 
    #> [[1]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.38 | 26.67, 30.08
    #>   2 |     24.21 | 22.71, 25.71
    #>   3 |     20.05 | 17.13, 22.96
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     31.82 | 28.07, 35.58
    #>   2 |     27.66 | 24.88, 30.43
    #>   3 |     23.49 | 20.63, 26.35
    #> 
    #> Adjusted for:
    #> * chl = 195.08
    #> 
    #> [[2]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.35 | 26.83, 31.87
    #>   2 |     26.01 | 23.56, 28.46
    #>   3 |     22.67 | 17.89, 27.44
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.35 | 23.58, 35.13
    #>   2 |     26.01 | 22.02, 30.00
    #>   3 |     22.67 | 18.73, 26.61
    #> 
    #> Adjusted for:
    #> * chl = 192.88
    #> 
    #> [[3]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     27.30 | 25.25, 29.35
    #>   2 |     25.40 | 23.54, 27.26
    #>   3 |     23.51 | 20.08, 26.94
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.87 | 24.24, 33.49
    #>   2 |     26.97 | 23.43, 30.52
    #>   3 |     25.08 | 21.50, 28.66
    #> 
    #> Adjusted for:
    #> * chl = 194.64
    #> 
    #> [[4]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     27.79 | 25.56, 30.03
    #>   2 |     24.66 | 22.52, 26.80
    #>   3 |     21.52 | 17.66, 25.38
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     31.87 | 27.10, 36.63
    #>   2 |     28.73 | 25.20, 32.26
    #>   3 |     25.59 | 22.01, 29.17
    #> 
    #> Adjusted for:
    #> * chl = 199.32
    #> 
    #> [[5]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.85 | 26.72, 30.99
    #>   2 |     24.47 | 22.75, 26.20
    #>   3 |     20.09 | 16.79, 23.40
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     31.72 | 27.13, 36.32
    #>   2 |     27.34 | 23.62, 31.07
    #>   3 |     22.96 | 18.94, 26.99
    #> 
    #> Adjusted for:
    #> * chl = 192.36
    #> 
    #> [[6]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.45 | 27.44, 31.45
    #>   2 |     25.67 | 23.80, 27.54
    #>   3 |     21.89 | 18.49, 25.30
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     30.64 | 26.49, 34.80
    #>   2 |     26.87 | 23.77, 29.96
    #>   3 |     23.09 | 19.84, 26.34
    #> 
    #> Adjusted for:
    #> * chl = 192.56
    #> 
    #> [[7]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.55 | 27.64, 31.46
    #>   2 |     25.64 | 23.98, 27.30
    #>   3 |     21.73 | 18.87, 24.58
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     30.48 | 27.60, 33.37
    #>   2 |     26.57 | 24.31, 28.84
    #>   3 |     22.66 | 19.79, 25.53
    #> 
    #> Adjusted for:
    #> * chl = 192.92
    #> 
    #> [[8]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.04 | 26.02, 30.07
    #>   2 |     25.29 | 23.62, 26.97
    #>   3 |     22.55 | 19.48, 25.62
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.09 | 25.15, 33.04
    #>   2 |     26.34 | 23.11, 29.58
    #>   3 |     23.60 | 19.96, 27.23
    #> 
    #> Adjusted for:
    #> * chl = 187.20
    #> 
    #> [[9]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.35 | 26.33, 30.37
    #>   2 |     25.06 | 22.85, 27.26
    #>   3 |     21.76 | 17.69, 25.84
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     30.93 | 26.65, 35.21
    #>   2 |     27.63 | 24.85, 30.41
    #>   3 |     24.34 | 21.49, 27.19
    #> 
    #> Adjusted for:
    #> * chl = 194.56
    #> 
    #> [[10]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.54 | 26.74, 30.34
    #>   2 |     25.15 | 23.50, 26.79
    #>   3 |     21.75 | 18.64, 24.87
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     32.04 | 27.92, 36.15
    #>   2 |     28.65 | 25.61, 31.68
    #>   3 |     25.25 | 22.25, 28.26
    #> 
    #> Adjusted for:
    #> * chl = 199.08
    

    Created on 2024-04-25 with reprex v2.1.0

    library(lspline)
    library(furrr)
    library(mice)
    library(ggeffects)
    
    imp <- mice(nhanes, print = FALSE, m = 10, seed = 24415)
    df <- complete(imp, "long")
    model_fit <- lapply(1:10, function(i) {
      model = lm(bmi ~ lspline(age, c(5, 10)) + hyp + chl,
                 data = subset(df, `.imp`==i))
    })
    
    plan(multisession, workers = 10)
    future_map(model_fit, ggpredict,  c("age[all]", "hyp"), progress = TRUE)
    #> [[1]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.38 | 26.67, 30.08
    #>   2 |     24.21 | 22.71, 25.71
    #>   3 |     20.05 | 17.13, 22.96
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     31.82 | 28.07, 35.58
    #>   2 |     27.66 | 24.88, 30.43
    #>   3 |     23.49 | 20.63, 26.35
    #> 
    #> Adjusted for:
    #> * chl = 195.08
    #> 
    #> [[2]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.35 | 26.83, 31.87
    #>   2 |     26.01 | 23.56, 28.46
    #>   3 |     22.67 | 17.89, 27.44
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.35 | 23.58, 35.13
    #>   2 |     26.01 | 22.02, 30.00
    #>   3 |     22.67 | 18.73, 26.61
    #> 
    #> Adjusted for:
    #> * chl = 192.88
    #> 
    #> [[3]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     27.30 | 25.25, 29.35
    #>   2 |     25.40 | 23.54, 27.26
    #>   3 |     23.51 | 20.08, 26.94
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.87 | 24.24, 33.49
    #>   2 |     26.97 | 23.43, 30.52
    #>   3 |     25.08 | 21.50, 28.66
    #> 
    #> Adjusted for:
    #> * chl = 194.64
    #> 
    #> [[4]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     27.79 | 25.56, 30.03
    #>   2 |     24.66 | 22.52, 26.80
    #>   3 |     21.52 | 17.66, 25.38
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     31.87 | 27.10, 36.63
    #>   2 |     28.73 | 25.20, 32.26
    #>   3 |     25.59 | 22.01, 29.17
    #> 
    #> Adjusted for:
    #> * chl = 199.32
    #> 
    #> [[5]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.85 | 26.72, 30.99
    #>   2 |     24.47 | 22.75, 26.20
    #>   3 |     20.09 | 16.79, 23.40
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     31.72 | 27.13, 36.32
    #>   2 |     27.34 | 23.62, 31.07
    #>   3 |     22.96 | 18.94, 26.99
    #> 
    #> Adjusted for:
    #> * chl = 192.36
    #> 
    #> [[6]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.45 | 27.44, 31.45
    #>   2 |     25.67 | 23.80, 27.54
    #>   3 |     21.89 | 18.49, 25.30
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     30.64 | 26.49, 34.80
    #>   2 |     26.87 | 23.77, 29.96
    #>   3 |     23.09 | 19.84, 26.34
    #> 
    #> Adjusted for:
    #> * chl = 192.56
    #> 
    #> [[7]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.55 | 27.64, 31.46
    #>   2 |     25.64 | 23.98, 27.30
    #>   3 |     21.73 | 18.87, 24.58
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     30.48 | 27.60, 33.37
    #>   2 |     26.57 | 24.31, 28.84
    #>   3 |     22.66 | 19.79, 25.53
    #> 
    #> Adjusted for:
    #> * chl = 192.92
    #> 
    #> [[8]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.04 | 26.02, 30.07
    #>   2 |     25.29 | 23.62, 26.97
    #>   3 |     22.55 | 19.48, 25.62
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     29.09 | 25.15, 33.04
    #>   2 |     26.34 | 23.11, 29.58
    #>   3 |     23.60 | 19.96, 27.23
    #> 
    #> Adjusted for:
    #> * chl = 187.20
    #> 
    #> [[9]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.35 | 26.33, 30.37
    #>   2 |     25.06 | 22.85, 27.26
    #>   3 |     21.76 | 17.69, 25.84
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     30.93 | 26.65, 35.21
    #>   2 |     27.63 | 24.85, 30.41
    #>   3 |     24.34 | 21.49, 27.19
    #> 
    #> Adjusted for:
    #> * chl = 194.56
    #> 
    #> [[10]]
    #> # Predicted values of bmi
    #> 
    #> hyp: 1
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     28.54 | 26.74, 30.34
    #>   2 |     25.15 | 23.50, 26.79
    #>   3 |     21.75 | 18.64, 24.87
    #> 
    #> hyp: 2
    #> 
    #> age | Predicted |       95% CI
    #> ------------------------------
    #>   1 |     32.04 | 27.92, 36.15
    #>   2 |     28.65 | 25.61, 31.68
    #>   3 |     25.25 | 22.25, 28.26
    #> 
    #> Adjusted for:
    #> * chl = 199.08
    

    Created on 2024-04-25 with reprex v2.1.0