@@ -4,11 +4,54 @@ Pre-1.0.0 numbering scheme: 0.x will indicate releases, while 0.x.y will indicat
44
55# epiprocess 0.9
66
7+ ## Breaking changes
8+
9+ - ` epi_slide ` interface has major breaking changes.
10+ - All variables are now dot-prefixed to be more consistent with tidyverse
11+ style for functions that allow tidyeval.
12+ - The ` before/after ` arguments have been replaced with the ` .window_size ` and
13+ ` .align ` arguments.
14+ - ` names_sep ` has been removed. If you return data frames from your
15+ computations:
16+ - without a name, they will be unpacked into separate columns without name
17+ prefixes
18+ - with a name, it will become a packed data.frame-class column (see
19+ ` tidyr::pack ` ).
20+ - ` as_list_col ` has been removed. You can now directly return a list from your
21+ slide computations instead. If you were using ` as_list_col=TRUE ` , you will
22+ need to wrap your output in a list.
23+ - Ungrouped slides are no longer allowed in ` epi_slide ` . If you used this for
24+ geographic aggregation up to national, consider using ` sum_groups_epi_df ` .
25+ - Added ` sum_groups_epi_df ` to allow aggregation across key columns prior to
26+ sliding.
27+ - ` epix_slide ` interface has major changes.
28+ - All variables are now dot-prefixed to be more consistent with tidyverse
29+ style for functions that allow tidyeval.
30+ - ` names_sep ` has been removed. If you return data frames from your
31+ computations:
32+ - without a name, they will be unpacked into separate columns without name
33+ prefixes
34+ - with a name, it will become a packed data.frame-class column (see
35+ ` tidyr::pack ` ).
36+ - ` as_list_col ` has been removed. You can now directly return a list from your
37+ slide computations instead. If you were using ` as_list_col=TRUE ` , you will
38+ need to wrap your output in a list.
39+ - ` as_epi_df() ` now checks that every group has unique time values and errors if
40+ this is not the case. The same check is performed at the beginning of
41+ ` epi_slide() ` . This check is currently not enforced in dplyr operations (like
42+ for joins, mutates, or select), but we plan to add it in the future.
43+ - ` as_epi_df() ` or ` as_epi_archive() ` no longer accept ` additional_metadata ` .
44+ Use the new ` other_keys ` arg to specify additional key columns, such as age
45+ group columns or other demographic breakdowns. Miscellaneous metadata are no
46+ longer handled by ` epiprocess ` , but you can use R's built-in ` attr<- ` instead
47+ for a similar feature.
48+
749## Improvements
850
951- Added ` complete.epi_df ` , which fills in missing values in an ` epi_df ` with
1052 ` NA ` s. Uses ` tidyr::complete ` underneath and preserves ` epi_df ` metadata.
11- - Inclusion of the function ` revision_summary ` to provide basic revision information for ` epi_archive ` s out of the box. (#492 )
53+ - Inclusion of the function ` revision_summary ` to provide basic revision
54+ information for ` epi_archive ` s out of the box. (#492 )
1255
1356## Bug fixes
1457
@@ -21,6 +64,14 @@ Pre-1.0.0 numbering scheme: 0.x will indicate releases, while 0.x.y will indicat
2164
2265## Breaking changes
2366
67+ - ` epi_df ` 's are now more strict about what types they allow in the time column.
68+ Namely, we are explicit about only supporting ` Date ` at the daily and weekly
69+ cadence and generic integer types (for yearly cadence).
70+ - ` epi_slide ` ` before ` and ` after ` arguments are now require the user to
71+ specific time units in certain cases. The ` time_step ` argument has been
72+ removed.
73+ - ` epix_slide ` ` before ` argument now defaults to ` Inf ` , and requires the user to
74+ specify units in some cases. The ` time_step ` argument has been removed.
2475- ` detect_outlr_stl(seasonal_period = NULL) ` is no longer accepted. Use
2576 ` detect_outlr_stl(seasonal_period = <value>, seasonal_as_residual = TRUE) `
2677 instead. See ` ?detect_outlr_stl ` for more details.
@@ -63,6 +114,12 @@ Pre-1.0.0 numbering scheme: 0.x will indicate releases, while 0.x.y will indicat
63114 are similar functions for ` geo ` and ` version ` ).
64115- Fixed bug where ` epix_slide_ref_time_values_default() ` on datetimes would
65116 output a huge number of ` ref_time_values ` spaced apart by mere seconds.
117+ - In ` epi_slide() ` and ` epix_slide() ` :
118+ - Multiple "data-masking" tidy evaluation expressions can be passed in via
119+ ` ... ` , rather than just one.
120+ - Additional tidy evaluation features from ` dplyr::mutate ` are supported: `!!
121+ name_var := value` , unnamed expressions evaluating to data frames, and ` =
122+ NULL` ; see ` ?epi_slide` for more details.
66123
67124## Cleanup
68125
@@ -71,17 +128,6 @@ Pre-1.0.0 numbering scheme: 0.x will indicate releases, while 0.x.y will indicat
71128 of ` epi_df ` s to let ` {epipredict} ` work more easily with other libraries (#471 ).
72129- Removed some external package dependencies.
73130
74- ## Breaking Changes
75-
76- - ` epi_df ` 's are now more strict about what types they allow in the time column.
77- Namely, we are explicit about only supporting ` Date ` at the daily and weekly
78- cadence and generic integer types (for yearly cadence).
79- - ` epi_slide ` ` before ` and ` after ` arguments are now require the user to
80- specific time units in certain cases. The ` time_step ` argument has been
81- removed.
82- - ` epix_slide ` ` before ` argument now defaults to ` Inf ` , and requires the user to
83- specify units in some cases. The ` time_step ` argument has been removed.
84-
85131# epiprocess 0.7.0
86132
87133## Breaking changes:
0 commit comments