Error: Subscript AMr1.orig is a matrix, the data x.imp[, -possibleFactors][AMr1.orig] must have size 1

[vincentarelbundock]

In case you had not seen this StackOverflow error report from 2020:

https://stackoverflow.com/questions/64056125/how-to-fix-erreur-subscript-amr1-orig-is-a-matrix-the-data-x-imp-possib

I just ran into the same problem. The solution is simple. Just add this to convert x to data frame if it is a tibble:

if (inherits(x, "tbl_df")) {
  x <- as.data.frame(x)
}

[vincentarelbundock]

FWIW, I ran into this issue with a complex dataset today. Hard to create a minimal reproducible example, but the SO question and answer have a lot of likes and comments, so I think the bug should be treated as "replicated on several machines."

[mecoskun]

Ran into this problem recently, and spent an hour trying to understand what was wrong. Fortunately, the SO post helped me resolve it. A bug fix would be helpful.

[raffaem]

Here is the most minimal reproducible example I can come up with:

#!/usr/bin/env R

# Load CHES
rm(list = ls())
library(tidyverse)
library(magrittr)
library(Amelia)

infp <- "ches.csv"
stopifnot(file.exists(infp))
ches <- readr::read_csv(infp, show_col_types = FALSE)

year_range <- ches %>%
  pull(year) %>%
  range()
years <- seq(year_range[1], year_range[2])

ches %<>% select(party_id, year, environment)
ches %<>% complete(party_id, year = years)
ches %<>% arrange(party_id, year)
#ches2 %<>% as.data.frame()
a.out <- amelia(ches, 
                m = 5, 
                ts = "year",
                cs = "party_id",
                polytime = 3)

ches.csv

Although the error message is slightly different:

Error in [<-:
! Subscript AMr1.orig is a matrix, the data x.imp[AMr1.orig] must have size 1.
Run rlang::last_trace() to see where the error occurred.

@IQSS any idea how to fix it?

[raffaem]

Maybe @mattblackwell ?