geocoder fail ?

[erikarasnick]

When I geocode the parsed_addresses, the first time I run through the geocoder, I get the error below. In the past, we have told people to just run it again, which works if you are able to utilize the geocoding_cache. So, I can get this to work if I don't use degauss_run but just use degauss at the command line, let it fail the first time, then run it again.

I'm guessing this is related to RAM? I have about 8 GB allocated to docker right now.

── Welcome to DeGAUSS! ──

• You are using geocoder, version 3.3.0
• This container returns geocodes
• <https://degauss.org/geocoder>

ℹ removing non-alphanumeric characters...
ℹ removing excess whitespace...
ℹ flagging PO boxes...
ℹ flagging known Cincinnati foster & institutional addresses...
ℹ flagging non-address text and missing addresses...
ℹ now geocoding ...
  |================================| 100%, Elapsed 01:01:51
16 errors occurred
the first one occurred in element 1225:

Error in readRDS(file = file.path(path, key)): error reading from connection

Error in `dplyr::mutate()`:
! Problem while computing `geocodes = purrr::map(geocodes, ~.x %>%
  purrr::map(unlist) %>% as_tibble())`.
Caused by error in `purrr::map()`:
ℹ In index: 1225.
Caused by error:
! Column 1 must be named.
Use .name_repair to specify repair.
Caused by error in `repaired_names()`:
! Names can't be empty.
✖ Empty name found at location 1.
Backtrace:
     ▆
  1. ├─... %>% dplyr::arrange(desc(precision), score)
  2. ├─dplyr::arrange(., desc(precision), score)
  3. ├─dplyr::mutate(...)
  4. ├─dplyr::select(., -fips_county, -prenum, -number, -row_index)
  5. ├─dplyr::rename(...)
  6. ├─dplyr::ungroup(.)
  7. ├─dplyr::slice(., 1)
  8. ├─dplyr::group_by(., row_index)
  9. ├─tidyr::unnest(., cols = c(geocodes))
 10. ├─dplyr::mutate(...)
 11. ├─dplyr:::mutate.data.frame(...)
 12. │ └─dplyr:::mutate_cols(.data, dplyr_quosures(...), caller_env = caller_env())
 13. │   ├─base::withCallingHandlers(...)
 14. │   └─mask$eval_all_mutate(quo)
 15. ├─purrr::map(geocodes, ~.x %>% purrr::map(unlist) %>% as_tibble())
 16. │ └─purrr:::map_("list", .x, .f, ..., .progress = .progress)
 17. │   ├─purrr:::with_indexed_errors(...)
 18. │   │ └─base::withCallingHandlers(...)
 19. │   └─.f(.x[[i]], ...)
 20. │     └─.x %>% purrr::map(unlist) %>% as_tibble()
 21. ├─tibble::as_tibble(.)
 22. └─tibble:::as_tibble.list(.)
 23.   └─tibble:::lst_to_tibble(x, .rows, .name_repair, col_lengths(x))
 24.     └─tibble:::set_repaired_names(x, repair_hint = TRUE, .name_repair)
 25.       ├─rlang::set_names(...)
 26.       └─tibble:::repaired_names(names2(x), repair_hint, .name_repair = .name_repair, quiet = quiet)
 27.         ├─tibble:::subclass_name_repair_errors(...)
 28.         │ └─base::withCallingHandlers(...)
 29.         └─vctrs::vec_as_names(...)
 30.           └─vctrs (local) `<fn>`()
 31.             └─vctrs:::validate_unique(names = names, arg = arg, call = call)
 32.               └─vctrs:::stop_names_cannot_be_empty(names, call = call)
 33.                 └─vctrs:::stop_names(...)
 34.                   └─vctrs:::stop_vctrs(...)
 35.                     └─rlang::abort(message, class = c(class, "vctrs_error"), ..., call = vctrs_error_call(call))
Execution halted

[erikarasnick]

The roads container also stopped part of the way through (proabably RAM again) and aadt finished and wrote output but then gave an error that vector memory was exhausted. I can try again with more RAM allocated to Docker...

also, I am trying this on my old Macbook, because the roads container "got stuck" due to arm64 emulation.

[cole-brokamp]

cole to verify

[cole-brokamp]

I've always been working with a subset of the data, so I haven't seen this problem yet. We will keep an eye out for when we scale up to the entire dataset.