[BUG]: Error in diagnostic plots

[chantelwetzel-noaa]

Is there an existing issue for this?

• [X] I have searched the existing issues

Current Behavior

The diagnostic plots error out for delta models (have not fully tested for tweedie models) on the following lines in the diagnose function:

    ignore <- purrr::map(
      seq_along(fit[["formula"]]),
      .f = function(x, y, f_dir) {
        y[["residuals"]] <- y[[
          paste0("residuals", ifelse(x == 1, "", x))
        ]]
        map_residuals(
          y,
          save_prefix = file.path(f_dir, paste0("residuals_", x, "_"))
        )
      },
      y = fit[["data"]],
      f_dir = dir
    )

The error that prints to the screen is:

And output from rlang::last_trace() is:

``` rlang::last_trace() Error in `dplyr::mutate()`: ℹ In argument: `results = purrr::pmap(...)`. Caused by error in `purrr::pmap()`: ℹ In index: 1. --- Backtrace: ▆ 1. ├─data %>% ... 2. ├─dplyr::mutate(...) 3. ├─dplyr:::mutate.data.frame(...) 4. │ └─dplyr:::mutate_cols(.data, dplyr_quosures(...), by) 5. │ ├─base::withCallingHandlers(...) 6. │ └─dplyr:::mutate_col(dots[[i]], data, mask, new_columns) 7. │ └─mask$eval_all_mutate(quo) 8. │ └─dplyr (local) eval() 9. ├─purrr::pmap(...) 10. │ └─purrr:::pmap_("list", .l, .f, ..., .progress = .progress) 11. │ ├─purrr:::with_indexed_errors(...) 12. │ │ └─base::withCallingHandlers(...) 13. │ ├─purrr:::call_with_cleanup(...) 14. │ └─indexwc (local) .f(...) 15. │ └─indexwc:::diagnose(dir = dir_index, fit = fit, prediction_grid = grid) at indexwc/R/run_sdmtmb.R:153:3 16. │ └─purrr::map(...) at indexwc/R/diagnose.R:82:5 17. │ └─purrr:::map_("list", .x, .f, ..., .progress = .progress) 18. │ ├─purrr:::with_indexed_errors(...) 19. │ │ └─base::withCallingHandlers(...) 20. │ ├─purrr:::call_with_cleanup(...) 21. │ └─indexwc (local) .f(.x[[i]], ...) 22. │ └─indexwc::map_residuals(...) at indexwc/R/diagnose.R:88:9 23. │ ├─purrr::map(...) at indexwc/R/map_residuals.R:63:3 24. │ │ └─purrr:::map_("list", .x, .f, ..., .progress = .progress) 25. │ │ └─purrr:::vctrs_vec_compat(.x, .purrr_user_env) 26. │ ├─base::seq(ggforce::n_pages(gg)) at indexwc/R/map_residuals.R:63:3 27. │ └─ggforce::n_pages(gg) at indexwc/R/map_residuals.R:63:3 28. │ ├─ggplot2::ggplot_build(plot) 29. │ └─ggplot2:::ggplot_build.ggplot(plot) 30. │ └─ggplot2:::by_layer(...) 31. │ ├─rlang::try_fetch(...) 32. │ │ ├─base::tryCatch(...) 33. │ │ │ └─base (local) tryCatchList(expr, classes, parentenv, handlers) 34. │ │ │ └─base (local) tryCatchOne(expr, names, parentenv, handlers[[1L]]) 35. │ │ │ └─base (local) doTryCatch(return(expr), name, parentenv, handler) 36. │ │ └─base::withCallingHandlers(...) 37. │ └─ggplot2 (local) f(l = layers[[i]], d = data[[i]]) 38. │ └─l$compute_aesthetics(d, plot) 39. │ └─ggplot2 (local) compute_aesthetics(..., self = self) 40. │ └─cli::cli_abort(...) 41. │ └─rlang::abort(...) 42. │ └─rlang:::signal_abort(cnd, .file) 43. │ └─base::signalCondition(cnd) 44. ├─rlang (local) ``(``) 45. │ └─handlers[[1L]](cnd) 46. │ └─cli::cli_abort(...) 47. │ └─rlang::abort(...) 48. │ └─rlang:::signal_abort(cnd, .file) 49. │ └─base::signalCondition(cnd) 50. ├─purrr (local) ``(``) 51. │ └─cli::cli_abort(...) 52. │ └─rlang::abort(...) 53. │ └─rlang:::signal_abort(cnd, .file) 54. │ └─base::signalCondition(cnd) 55. └─purrr (local) ``(``) 56. └─cli::cli_abort(...) --- Caused by error in `purrr::map()` at indexwc/R/diagnose.R:82:5: ℹ In index: 2. Caused by error in `ggplot2::geom_point()` at indexwc/R/map_residuals.R:41:3: ! Problem while computing aesthetics. ℹ Error occurred in the 2nd layer. --- Backtrace: ▆ 1. ├─data %>% ... 2. ├─dplyr::mutate(...) 3. ├─dplyr:::mutate.data.frame(...) 4. │ └─dplyr:::mutate_cols(.data, dplyr_quosures(...), by) 5. │ ├─base::withCallingHandlers(...) 6. │ └─dplyr:::mutate_col(dots[[i]], data, mask, new_columns) 7. │ └─mask$eval_all_mutate(quo) 8. │ └─dplyr (local) eval() 9. ├─purrr::pmap(...) 10. │ └─purrr:::pmap_("list", .l, .f, ..., .progress = .progress) 11. │ ├─purrr:::with_indexed_errors(...) 12. │ │ └─base::withCallingHandlers(...) 13. │ ├─purrr:::call_with_cleanup(...) 14. │ └─indexwc (local) .f(...) 15. │ └─indexwc:::diagnose(dir = dir_index, fit = fit, prediction_grid = grid) at indexwc/R/run_sdmtmb.R:153:3 16. │ └─purrr::map(...) at indexwc/R/diagnose.R:82:5 17. │ └─purrr:::map_("list", .x, .f, ..., .progress = .progress) 18. │ ├─purrr:::with_indexed_errors(...) 19. │ │ └─base::withCallingHandlers(...) 20. │ ├─purrr:::call_with_cleanup(...) 21. │ └─indexwc (local) .f(.x[[i]], ...) 22. │ └─indexwc::map_residuals(...) at indexwc/R/diagnose.R:88:9 23. │ ├─purrr::map(...) at indexwc/R/map_residuals.R:63:3 24. │ │ └─purrr:::map_("list", .x, .f, ..., .progress = .progress) 25. │ │ └─purrr:::vctrs_vec_compat(.x, .purrr_user_env) 26. │ ├─base::seq(ggforce::n_pages(gg)) at indexwc/R/map_residuals.R:63:3 27. │ └─ggforce::n_pages(gg) at indexwc/R/map_residuals.R:63:3 28. │ ├─ggplot2::ggplot_build(plot) 29. │ └─ggplot2:::ggplot_build.ggplot(plot) 30. │ └─ggplot2:::by_layer(...) 31. │ ├─rlang::try_fetch(...) 32. │ │ ├─base::tryCatch(...) 33. │ │ │ └─base (local) tryCatchList(expr, classes, parentenv, handlers) 34. │ │ │ └─base (local) tryCatchOne(expr, names, parentenv, handlers[[1L]]) 35. │ │ │ └─base (local) doTryCatch(return(expr), name, parentenv, handler) 36. │ │ └─base::withCallingHandlers(...) 37. │ └─ggplot2 (local) f(l = layers[[i]], d = data[[i]]) 38. │ └─l$compute_aesthetics(d, plot) 39. │ └─ggplot2 (local) compute_aesthetics(..., self = self) 40. │ └─cli::cli_abort(...) 41. │ └─rlang::abort(...) 42. │ └─rlang:::signal_abort(cnd, .file) 43. │ └─base::signalCondition(cnd) 44. └─rlang (local) ``(``) 45. └─handlers[[1L]](cnd) 46. └─cli::cli_abort(...) Caused by error in `compute_aesthetics()`: ! Aesthetics are not valid data columns. ✖ The following aesthetics are invalid: ✖ `colour = NULL` ✖ `fill = NULL` ℹ Did you mistype the name of a data column or forget to add `after_stat()`? --- Backtrace: ▆ 1. ├─data %>% ... 2. ├─dplyr::mutate(...) 3. ├─dplyr:::mutate.data.frame(...) 4. │ └─dplyr:::mutate_cols(.data, dplyr_quosures(...), by) 5. │ ├─base::withCallingHandlers(...) 6. │ └─dplyr:::mutate_col(dots[[i]], data, mask, new_columns) 7. │ └─mask$eval_all_mutate(quo) 8. │ └─dplyr (local) eval() 9. └─purrr::pmap(...) 10. └─purrr:::pmap_("list", .l, .f, ..., .progress = .progress) 11. ├─purrr:::with_indexed_errors(...) 12. │ └─base::withCallingHandlers(...) 13. ├─purrr:::call_with_cleanup(...) 14. └─indexwc (local) .f(...) 15. └─indexwc:::diagnose(dir = dir_index, fit = fit, prediction_grid = grid) at indexwc/R/run_sdmtmb.R:153:3 16. └─purrr::map(...) at indexwc/R/diagnose.R:82:5 17. └─purrr:::map_("list", .x, .f, ..., .progress = .progress) 18. ├─purrr:::with_indexed_errors(...) 19. │ └─base::withCallingHandlers(...) 20. ├─purrr:::call_with_cleanup(...) 21. └─indexwc (local) .f(.x[[i]], ...) 22. └─indexwc::map_residuals(...) at indexwc/R/diagnose.R:88:9 23. ├─purrr::map(...) at indexwc/R/map_residuals.R:63:3 24. │ └─purrr:::map_("list", .x, .f, ..., .progress = .progress) 25. │ └─purrr:::vctrs_vec_compat(.x, .purrr_user_env) 26. ├─base::seq(ggforce::n_pages(gg)) at indexwc/R/map_residuals.R:63:3 27. └─ggforce::n_pages(gg) at indexwc/R/map_residuals.R:63:3 28. ├─ggplot2::ggplot_build(plot) 29. └─ggplot2:::ggplot_build.ggplot(plot) 30. └─ggplot2:::by_layer(...) 31. ├─rlang::try_fetch(...) 32. │ ├─base::tryCatch(...) 33. │ │ └─base (local) tryCatchList(expr, classes, parentenv, handlers) 34. │ │ └─base (local) tryCatchOne(expr, names, parentenv, handlers[[1L]]) 35. │ │ └─base (local) doTryCatch(return(expr), name, parentenv, handler) 36. │ └─base::withCallingHandlers(...) 37. └─ggplot2 (local) f(l = layers[[i]], d = data[[i]]) 38. └─l$compute_aesthetics(d, plot) 39. └─ggplot2 (local) compute_aesthetics(..., self = self) Run rlang::last_trace(drop = FALSE) to see 5 hidden frames. ``` ### Expected Behavior _No response_ ### Steps To Reproduce The model was run using the template call in the configuration.R file where the call to the model is: ``` to_run <- "aurora rockfish" species_data <- catch[which(catch$Common_name %in% to_run), ] configuration <- tibble::as_tibble(read.csv( file.path("data-raw", "configuration.csv") )) configuration <- configuration[which(configuration$species %in% to_run), ] data <-configuration %>% # Row by row ... do stuff then ungroup dplyr::rowwise() %>% # Pull the data based on the function found in fxn column dplyr::mutate( data_raw = list(format_data(data = species_data)), data_filtered = list(data_raw %>% dplyr::filter( depth <= min_depth, depth >= max_depth, latitude >= min_latitude, latitude <= max_latitude, year >= min_year, year <= max_year )) ) %>% dplyr::ungroup() best <- data %>% dplyr::mutate( # Evaluate the call in family family = purrr::map(family, .f = ~ eval(parse(text = .x))), # Run the model on each row in data results = purrr::pmap( .l = list( data = data_filtered, formula = formula, family = family, anisotropy = anisotropy, n_knots = knots, spatiotemporal = purrr::map2(spatiotemporal1, spatiotemporal2, list) ), .f = indexwc::run_sdmtmb ) ) ```

Environment

Encountered using R version 4.2.2

Anything else?

No response

[chantelwetzel-noaa]

Attaching a txt document with the output from rlang::last_trace() since the format got messed up in the original issue.

diagnose_error.txt

[kellijohnson-NOAA]

@chantelwetzel-noaa I have been doing some research to try and remedy this bug that I built into the code accidently. And, I believe that the easiest way to fix it is to change the configuration file to just have a single column for spatiotemporal instead of spatiotemporal1 and spatiotemporal2. That way we can pass either a vector or a single character to the spatiotemporal column. Does that sound like an okay fix? And, sorry it took me so long to wrap my head around how to get this done.

[chantelwetzel-noaa]

I think that proposed approach is reasonable and would resolve the issues I was encountering when running tweedie model (see fix-tweedie branch) but I am not sure if that would fix this error. Let me test this out locally and let you know what I get back.

[kellijohnson-NOAA]

Right, I got mixed up on the issue. This is an error with the mapping of the residuals. I just got the same error on my machine, so I can investigate it.

[kellijohnson-NOAA]

This is happening because in fit[["data"]] the names are as follows:

Browse[1]> names(fit[["data"]])
 [1] "year"                 "fyear"                "survey_name"
 [4] "common_name"          "catch_numbers"        "catch_weight"
 [7] "effort"               "pass_scaled"          "vessel_year"
[10] "longitude"            "latitude"             "depth"
[13] "depth_scaled"         "depth_scaled_squared" "x"
[16] "y"                    "residuals"

and it is expected that there would be "residuals_2" as well.

[kellijohnson-NOAA]

Looks like I broke the code here:

  if (
    lookup_is_delta(fit[["formula"]]) &&
    lookup_is_mixture(fit[["formula"]])
  ) {
    fit[["data"]][["residuals2"]] <- stats::residuals(fit, model = 2)
  }

I will fix it.