New function - overlap_method_names(). Overlap methods for a corresponding overlap method codes.
"chain" overlap method split into "x_chain_y" and "y_chain_x". "chain" will continue to be supported as a keyword for "x_chain_y" OR "y_chain_x" method"across" overlap method split into "x_across_y" and "y_across_x". "across" will continue to be supported as a keyword for "x_across_y" OR "y_across_x" methods"inbetween" overlap method split into "x_inbetween_y" and "y_inbetween_x". "inbetween" will continue to be supported as a keyword for "x_inbetween_y" OR "y_inbetween_x" methodsoverlaps().overlap_method_names().make_batch_pairs() (internal) created invalid record pairs. Resolved.reframe(). Modify the attributes of a sub_criteria object.link_records(). Record linkage by creating all record pairs as opposed to batches as with link().make_pairs(). Create every combination of records-pairs for a given dataset.make_pairs_wf_source(). Create records-pairs from different sources only.make_ids(). Convert an edge list to a group identifier.merge_ids(). Merge two group identifiers.attrs(). Pass a set of attributes to one instance of match_funcs or equal_funcs.episodes_wf_splits()episodes() and links(). Reduced processing times.display argument. "progress_with_report", "stats_with_report" and "none_with_report". Creates a d_report; a status of the analysis over its run time.eval_sub_criteria(). Record-pairs are no longer created in the function. Therefore, index_record and sn arguments have been replaced with x_pos and y_pos.link_records() and links_wf_probabilistic(). The cmp_threshold argument has been renamed to attr_threshold.show_labels argument in schema(). Two new options - "wind_nm" and "length" to replace "length_label".wind_id list in episodes(..., data_link = "XX") in . Resolved.link_id in links(..., recursive = TRUE). Resolved.iteration not recorded in some situations with episodes(). Resolved.skip_order ends an open episode. Resolved.NA in dist_wind_index and dist_epid_index when sn is supplied. Resolved.overlap_method_codes() - overlap method codes not recycled properly. Resolved.delink(). Unlink identifiers.episodes_wf_splits(). Wrapper function of episodes(). Better optimised for handling datasets with many duplicate records.combi(). Numeric codes for unique combination of vectors.attr_eval(). Recursive evaluation of a function on each attribute of a sub_criteria.case_nm values - Case_CR and Recurrence_CR which are Case and Recurrence without a sub-criteria match.schema.epid.eval_sub_criteria with 1 result.links_wf_probabilistic(). Probabilistic record linkage.partitions(). Spilt events into sections in time.schema(). Plot schema diagrams for pid, epid, pane and number_line objects.encode() and decode(). Encode and decode slots values to minimise memory usage.episodes() - case_sub_criteria and recurrence_sub_criteria. Additional matching conditions for temporal links.episodes()- case_length_total and recurrence_length_total. Number of temporal links required for a window/episode.links() - recursive. Control if matches can spawn new matches.links() - check_duplicates. Control the checking of logical tests on duplicate values. If FALSE, results are recycled for the duplicates.as.data.frame and as.list S3 methods for the pid, number_line, epid, pane objects.episode_type in episodes() - “recursive”. For recursive episodes where every linked events can be used as a subsequent index event.recurrence_from_last renamed to reference_event and given two new options.episodes() and links(). Speed improvements.epid_interval or pane_interval with POSIXct objects is now “GMT”.number_line_sequence() - splits number_line objects. Also available as a seq method.epid_total, pid_total and pane_total slots are populated by default. No need to used group_stats to get these.to_df() - Removed. Use as.data.frame() instead.to_s4() - Now an internal function. It’s no longer exported.compress_number_line() - Now an internal function. It’s no longer exported. Use episodes() instead.sub_criteria() - produces a sub_criteria object. Nested “AND” and “OR” conditions are now possible.case_overlap_methods, recurrence_overlap_methods and overlap_methods now take integer codes for different combinations of overlap methods. See overlap_methods$options for the full list. character inputs are still supported."Single-record" was wrong in links summary output. Resolved.Inf in number_line objects.case_length or recurrence_length for the same event.
overlap_methods for the corresponding case_length and recurrence_length.links() to replace record_group().sub_criteria(). The new way of supplying a sub_criteria in links().exact_match(), range_match() and range_match_legacy(). Predefined logical tests for use with sub_criteria(). User-defined tests can also be used. See ?sub_criteria.custom_sort() for nested sorting.epid_lengths() to show the required case_length or recurrence_length for an analyses. Useful in confirming the required case_length or recurrence_length for episode tracking.epid_windows(). Shows the period a date will overlap with given a particular case_length or recurrence_length. Useful in confirming the required case_length or recurrence_length for episode tracking.strata in links(). Useful for stratified data linkage. As in stratified episode tracking, a record with a missing strata (NA_character_) is skipped from data linkage.data_links in links(). Unlink record groups that do not include records from certain data sourceslistr(). Format atomic vectors as a written list.combns(). An extension of combn to generate permutations not ordinarily captured by combn.iteration slot for pid and epid objectsoverlap_method - reverse()number_line() - l and r must have the same length or be 1.episodes() - case_nm differentiates between duplicates of "Case" ("Duplicate_C") and "Recurrent" events ("Duplicate_R").episodes().
"Case").
episode_type - simultaneously track both "fixed" and "rolling" episodes.skip_if_b4_lengths - simultaneously track episodes where events before a cut-off range are both skipped and not skipped.episode_unit - simultaneously track episodes by different units of time.case_for_recurrence - simultaneously track "rolling" episodes with and without an additional case window for recurrent events.recurrence_from_last - simultaneously track "rolling" episodes with reference windows calculated from the first and last event of the previous window.strata. Options must be the same in each strata.
from_last - simultaneously track episodes in both directions of time - past to present and present to past.episodes_max - simultaneously track different number of episodes within the dataset.include_overlap_method - "overlap" and "none" will not be combined with other methods.
"overlap" - mutually inclusive with the other methods, so their inclusion is not necessary."none" - mutually exclusive and prioritised over the other methods (including "none"), so their inclusion is not necessary.NA_real_) or periods (number_line(NA_real_, NA_real_)) case_length and recurrence_length. This ensures that the event does not become an index case however, it can still be part of different episode. For reference, an event with a missing strata (NA_character_) ensures that the event does not become an index case nor part of any episode.fixed_episodes, rolling_episodes and episode_group - include_index_period didn’t work in certain situations. Corrected.fixed_episodes, rolling_episodes and episode_group - dist_from_wind was wrong in certain situations. Corrected.record_group() - strata. Perform record linkage separately within subsets of a dataset.overlap(), compress_number_line(), fixed_sepisodes(), rolling_episodes() and episode_group() - overlap_methods and methods. Replaces overlap_method and method respectively. Use different sets of methods within the same dataset when grouping episodes or collapsing number_line objects. overlap_method and method only permits 1 method per per dataset.epid objects - win_nm. Shows the type of window each event belongs to i.e. case or recurrence windowepid objects - win_id. Unique ID for each window. The ID is the sn of the reference event for each window
epid objects updated to reflect thisepid objects - dist_from_wind. Shows the duration of each event from its window’s reference eventepid objects - dist_from_epid. Shows the duration of each event from its episode’s reference eventepisode_group() and rolling_episodes() - recurrence_from_last. Determine if reference events should be the first or last event from the previous window.episode_group() and rolling_episodes() - case_for_recurrence. Determine if recurrent events should have their own case windows or not.episode_group(), fixed_episodes() and rolling_episodes() - data_links. Unlink episodes that do not include records from certain data_source(s).episode_group(), fixed_episodes() and rolling_episodes() - case_length and recurrence_length arguments. You can now use a range (number_line object).episode_group(), fixed_episodes() and rolling_episodes() - include_index_period. If TRUE, overlaps with the index event or period are grouped together even if they are outside the cut-off range (case_length or recurrence_length).pid objects - link_id. Shows the record (sn slot) to which every record in the dataset has matched to.invert_number_line(). Invert the left and/or right points to the opposite end of the number lineleft_point(x)<-, right_point(x)<-, start_point(x)<- and end_point(x)<-overlap() renamed to overlaps(). overlap() is now a convenience overlap_method to capture ANY kind of overlap."none" is another convenience overlap_method for NO kind of overlapexpand_number_line() - new options for point; "left" and "right"compress_number_line() - compressed number_line object inherits the direction of the widest number_line among overlapping group of number_line objectsoverlap_methods - have been changed such that each pair of number_line objects can only overlap in one way. E.g.
"chain" and "aligns_end" used to be possible but this is now considered a "chain" overlap only"aligns_start" and "aligns_end" use to be possible but this is now considered an "exact" overlapnumber_line_sequence() - Output is now a list.number_line_sequence() - now works across multiple number_line objects.to_df() - can now change number_line objects to data.frames.
to_s4() can do the reverse.epid objects are the default outputs for fixed_episodes(), rolling_episodes() and episode_group()pid objects are the default outputs for record_group()case_nm for events that were skipped due to rolls_max or episodes_max is now "Skipped".episode_group() and record_group(), sn can be negative numbers but must still be uniqueepisode_group() and record_group(). Runs just a little bit faster …x and y to have the same lengths in overlap functions.
episode_group - case_length and recurrence_length arguments. Now accepts negative numbers.
end_point() of the first period.
number_line_width(), both will be collapsed if the second one is within some days (or any other episode_unit) before the start_point() of the first period.case_nm wasn’t right for rolling episodes. Resolvedepisode_group(), fixed_episodes() and rolling_episodes() - optimized to take less time when working with large datasetsepisode_group(), fixed_episodes() and rolling_episodes() - date argument now supports numeric valuescompress_number_line() - the output (gid slot) is now a group identifier just like in epid objects (epid_interval)pid S4 object class for results of record_group(). This will replace the current default (data.frame) in the next major releaseepid S4 object class for results of episode_group(), fixed_episodes() and rolling_episodes(). This will replace the current default (data.frame) in the next releaseto_s4() and to_s4 argument in record_group(), episode_group(), fixed_episodes() and rolling_episodes(). Changes their output from a data.frame (current default) to epid or pid objectsto_df() changes epid or pid objects to a data.framededuplicate argument from fixed_episodes() and rolling_episodes() added to episode_group()fixed_episodes() and rolling_episodes() are now wrapper functions of episode_group(). Functionality remains the same but now includes all arguments available to episode_group()fixed_episodes() and rolling_episodes() from number_line to data.frame, pending the change to epid objectspid_cri column returned in record_group is now numeric. 0 indicates no match.criteria multiple times record_group()number_line objects can now be used as a criteria in record_group()episode_unit in episode_group()bi_direction in episode_group()fixed_episodes() and rolling_episodes() - Group records into fixed or rolling episodes of events or period of events.episode_group() - A more comprehensive implementation of fixed_episodes() and rolling_episodes(), with additional features such as user defined case assignment.record_group() - Multistage deterministic linkage that addresses missing data.number_line S4 object.
record_group()fixed_episodes(), rolling_episodes() and episode_group()fixed_episodes() and rolling_episodes()