jamovi

Hello,

I am just starting to develop modules for Jamovi and have successfully implemented a few so far. However, I am facing an issue with one specific module, and I would greatly appreciate some assistance.

The module I am working on is for data imputation using the missRanger package.

Everything seems to be working fine, but I keep receiving the message:
"This analysis has terminated, likely due to hitting a resource limit."

Below are my code files for your review.

Would anyone have any ideas on what might be causing this issue?

.a.yaml File:

name: missRangerImputation
title: MissRanger Imputation
menuGroup: BSS
menuSubgroup: Data handling
menuSubtitle: Under development...
version: '0.0.1'
jas: '1.2'

options:
- name: data
type: Data

- name: vars
title: Variables to Impute
type: Variables
suggested:
- continuous
- nominal
- ordinal
required: true
permitted:
- numeric
- factor

- name: maxiter
title: Maximum Iterations
type: Integer
default: 10
min: 1
max: 100

- name: num_trees
title: Number of Trees
type: Integer
default: 100
min: 1
max: 500

- name: pmm_k
title: Number of Donors for PMM
type: Integer
default: 5
min: 1
max: 20

- name: saveData
title: Add Imputed Values as New Column
type: Bool
default: false

.r.yaml File:

name: missRangerImputation
title: MissRanger Imputation Results
jrs: '1.1'

items:
- name: text
title: MissRanger Imputation Summary
type: Preformatted

- name: missingSummary
title: Missing Data Summary
type: Table
rows: 0
columns:
- name: variable
title: Variable
type: text
- name: total_missing
title: Missing Values
type: integer
- name: percent_missing
title: Missing (%)
type: number
format: '%.2f'

- name: imputedDataTable
title: Imputed Data Summary
type: Table
rows: 0
columns:
- name: variable
title: Variable
type: text
- name: original_value
title: Original Value
type: text
- name: imputed_value
title: Imputed Value
type: text
- name: row_number
title: Row
type: integer


.b.R File:

missRangerImputationClass <- if (requireNamespace('R6', quietly=TRUE)) R6::R6Class(
"missRangerImputationClass",
inherit = missRangerImputationBase,
public = list(
initialize = function(...) {
super$initialize(...)
},
run = function() {
# Checks if variables are selected
if (length(self$options$vars) == 0) {
self$results$text$setContent("Please select at least one variable for imputation.")
return()
}

# Checks if the 'missRanger' package is available
if (!requireNamespace("missRanger", quietly = TRUE)) {
stop("The 'missRanger' package is not installed.")
}

# Loads the package
library(missRanger)

# Prepares the dataset
df <- self$data
selected_vars <- self$options$vars

# Creates missing data summary
for (var in selected_vars) {
n_missing <- sum(is.na(df[[var]]))
pct_missing <- (n_missing / nrow(df)) * 100

self$results$missingSummary$addRow(list(
variable = var,
total_missing = n_missing,
percent_missing = pct_missing
))
}

# If no missing data, stops execution
missing_summary_df <- self$results$missingSummary$asDF()

if (nrow(missing_summary_df) == 0 || sum(missing_summary_df$total_missing) == 0) {
self$results$text$setContent("None of the selected variables have missing values for imputation.")
return()
}

# Performs imputation
imputed_data <- tryCatch({
message("Starting imputation...")
missRanger(
data = df,
maxiter = self$options$maxiter,
num.trees = self$options$num_trees,
pmm.k = self$options$pmm_k,
verbose = TRUE
)
}, error = function(e) {
self$results$text$setContent(paste("Error during imputation:", e$message))
return(NULL)
})

if (is.null(imputed_data)) return()

# Adds imputed data to results table
for (var in selected_vars) {
na_indices <- which(is.na(df[[var]]))

if (length(na_indices) > 0) {
for (idx in na_indices) {
original_val <- df[[var]][idx]
imputed_val <- imputed_data[[var]][idx]

self$results$imputedDataTable$addRow(list(
variable = var,
original_value = "NA",
imputed_value = as.character(imputed_val),
row_number = idx
))
}
}
}

# Updates text summary
total_imputations <- nrow(self$results$imputedDataTable$asDF())
summary_text <- sprintf(
"Imputation completed!\n\nDetails:\n- Number of trees: %d\n- Max iterations: %d\n- PMM donors: %d\n- Imputed variables: %s\n- Total imputations: %d",
self$options$num_trees,
self$options$maxiter,
self$options$pmm_k,
paste(selected_vars, collapse = ", "),
total_imputations
)
self$results$text$setContent(summary_text)

# Adds imputed data as new columns if requested
if (self$options$saveData) {
for (var in selected_vars) {
new_col_name <- paste0(var, "_imputed")
self$data[[new_col_name]] <- imputed_data[[var]]
}
}
}
)
)


Thanks in advance for any suggestions!

hi,

the error "This analysis has terminated, likely due to hitting a resource limit." means that the R process running the analysis has crashed.

you could try running jamovi at the terminal, and see if there's additional error messages.

under macOS this would be running

/Applications/jamovi.app/Contents/MacOS/jamovi

under windows you'll find "jamovi debug" in the start menu, and you can start jamovi with

jamovi --debug2

(the positions of the dashes and spaces isn't negotiable).

when run like this under windows, you'll see some additional windows are created (one for each engine process), and when an engine crashes, error ouput is typically written to one of the windows... but this is where it gets tricky, because after the crash, the window disappears before you can really read what it says. i normally end up running screen recording software, and then step through the recording frame by frame so i can read the error text. bit of a pita.

jonathon

Hello,

Thank you so much for the quick response and the helpful information.

I followed your instructions, and I encountered the following error:

Error in (function ( ) : unused argument (noThrow = TRUE)
terminate called after throwing an instance of 'Rcpp::LongjumpException'

(image URL) https://drive.google.com/file/d/1inFGt- ... drive_link


Another error also appeared, but I am not sure if it is significant. Regardless, I have attached an image below for your review.

Engine process terminated Exit code: 3221225786
(image URL) https://drive.google.com/file/d/1KdKY3c ... drive_link

Thank you again for your help!

Rafael.

ah, i think the issue is because you have a function called run() in your module ... this is clobbering the run() method in the analysis class. i think you want to call it .run() rather than run()

cheers

jonathon

Thank you so much! I applied your suggested fix, and the error disappeared.

However, the table is not populating as expected, so I realize I still need to refine my code further.

I appreciate your help and guidance—thank you again!

Regards,

Rafael.

I hope you are doing well.

I am reaching out to inquire about any progress regarding the development of a Jamovi module for missing data imputation using the missRanger R package. Given the critical role of handling missing data in statistical analysis, integrating this functionality would be a valuable addition to the Jamovi ecosystem.

Looking forward to your response.

Best regards,

Good morning,

I would also like to ask you about the progress in the development of a Jamovi imputation module using the missRanger R package. I have several projects that require the use of multiple imputation, and I would certainly prefer to do this in Jamovi. Thank you!

Hi

You can use jYS module in jamovi library now.

Warm Regards
Seol

You have interested me, I will try to add missRanger to jYS module as an second alternative to missForest

A heartfelt thank you to yurismol for their initiative to integrate the missRanger package into jamovi, following their development of the module based on missForest. This addition significantly enhances the imputation options available to our users.

missRanger, while inspired by the missForest algorithm, offers notable advantages:

Increased speed through the use of the {ranger} package, enabling more efficient imputation, especially on large datasets.

Incorporation of Predictive Mean Matching (PMM), which prevents the imputation of values not present in the original data and improves the variance of conditional distributions, crucial for multiple imputations .

Support for out-of-sample applications, providing additional flexibility in analyses .


This advancement strengthens the robustness and flexibility of imputation tools in jamovi. Once again, thank you for this valuable contribution!