Bringing your own data
You can use any surveillance dataset with acciddasuite —
just pass it through check_data() to validate and enter the
pipeline.
Required columns
Your data frame must have these 4 columns:
| Column | Type | Description |
|---|---|---|
target_end_date |
Date | The date for which an observation is recorded |
observation |
numeric | The observed value |
location |
character | A single location identifier |
target |
character | A single target identifier (e.g., “inc hosp influenza”) |
To enable nowcasting (correcting for reporting delays), add a 5th column:
| Column | Type | Description |
|---|---|---|
as_of |
Date | The date the observation was reported/revised |
With as_of, the same target_end_date can
appear multiple times (one row per revision). Without it, each
target_end_date should appear once.
Example
library(acciddasuite)
head(df)## target_end_date observation location target
## 1 2024-01-01 13 NY inc hosp influenza
## 2 2024-01-08 15 NY inc hosp influenza
## 3 2024-01-15 19 NY inc hosp influenza
## 4 2024-01-22 22 NY inc hosp influenza
## 5 2024-01-29 25 NY inc hosp influenza
## 6 2024-02-05 11 NY inc hosp influenza
checked <- check_data(df)
checked## <accidda_data>
##
## Location: NY
## Target: inc hosp influenza
## Window: 2024-01-01 to 2024-12-23 ( 52 dates )
## History: FALSEThe check_data() output tells you whether revision
history is available. From here you can go directly to forecasting:
get_fcast(checked, eval_start_date = "2024-11-01", h = 4)Or, if your data has revision history, nowcast first: