I attach (at bottom) two example submission files and two example tasks.json config files adapted from the set up in FluSight. In both files, a quantile level of 0.1 is represented in the file as 0.1000000000000000055511, which is the floating point representation of that value (at least, on my machine). The difference is whether the hub additionally includes another target with pmf output_type, where the output_type_id is a character string.
When I run validations for these output files, I get the following results:
Using the tasks.json file that includes a character output_type:
For the submission file that includes the pmf outputs, I see the following output:
Using the tasks.json file that does not include a character output type
However, for the submission file that does not include the pmf outputs, I get the following output with no errors related to this check:
+ file_path="UMass-gbq_bootstrap/2023-11-04-UMass-gbq_bootstrap.csv")
✔ FluSight-forecast-hub: All hub config files are valid.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: File exists at path
model-output/UMass-gbq_bootstrap/2023-11-04-UMass-gbq_bootstrap.csv.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: File name "2023-11-04-UMass-gbq_bootstrap.csv" is valid.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: File directory name matches `model_id` metadata in file name.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: `round_id` is valid.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: File is accepted hub format.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: Metadata file exists at path model-metadata/UMass-gbq_bootstrap.yml.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: File could be read successfully.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: `round_id_col` name is valid.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: `round_id` column "reference_date" contains a single, unique round ID
value.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: All `round_id_col` "reference_date" values match submission `round_id`
from file name.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: Column names are consistent with expected round task IDs and std
column names.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: Column data types match hub schema.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: `tbl` contains valid values/value combinations.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: All combinations of task ID column/`output_type`/`output_type_id`
values are unique.
! 2023-11-04-UMass-gbq_bootstrap.csv: Required task ID/output type/output type ID combinations missing. See
`missing` attribute for details.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: Values in column `value` all valid with respect to modeling task
config.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: Values in `value` column are non-decreasing as output_type_ids
increase for all unique task ID value/output type combinations of quantile or cdf output types.
ℹ 2023-11-04-UMass-gbq_bootstrap.csv: No pmf output types to check for sum of 1. Check skipped.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: Time differences between t0 var `reference_date` and t1 var
`target_end_date` all match expected period of 7d 0H 0M 0S * `horizon`.
✔ 2023-11-04-UMass-gbq_bootstrap.csv: Target counts are less than location population size.
! 2023-11-04-UMass-gbq_bootstrap.csv: Submission time must be within accepted submission window for round.
Current time 2023-10-21 18:15:30 is outside window 2023-10-29 EDT--2023-11-01 23:59:59 EDT.
Summary
Ideally, whether or not this submission file was valid would not depend on configuration settings for other targets.
I attach (at bottom) two example submission files and two example
tasks.json
config files adapted from the set up in FluSight. In both files, a quantile level of0.1
is represented in the file as0.1000000000000000055511
, which is the floating point representation of that value (at least, on my machine). The difference is whether the hub additionally includes another target with pmfoutput_type
, where theoutput_type_id
is a character string.When I run validations for these output files, I get the following results:
Using the tasks.json file that includes a character output_type:
For the submission file that includes the pmf outputs, I see the following output:
For the submission file that does not include the pmf outputs, I get a similar result:
Using the tasks.json file that does not include a character output type
However, for the submission file that does not include the pmf outputs, I get the following output with no errors related to this check:
Summary
Ideally, whether or not this submission file was valid would not depend on configuration settings for other targets.
tasks_with_char_output_type.json tasks_without_char_output_type.json 2023-10-28-UMass-gbq_bootstrap.csv 2023-11-04-UMass-gbq_bootstrap.csv