Changes in this release:
Test for new ks.test behavior in r-devel. (@brews, PR #191)
Update DOI badge and support documentation on main website and README. (@chguiterman, PR #182)
Changes in this release:
Add rlang, stringr, forcats, and tidyr as package dependencies.
Fix bug in plot_demograph() to handle discontinuous recording periods. Thanks to Mannie Lopez for finding that this addition disrupted the series sorting. (@chguiterman, #78)
Fix bug where "springdormant_fs" was omitted from the list of scar types in burnr:::rec_type_recording. This caused problems in the calculation of percent_scarred() because too few trees were recognized as recording in the year of fire. This would have also caused problems in composite(). Thanks to @GuobaoXu for finding the error and tracing its origin. (@chguiterman, #162, #167)
Change to format of FHX2 files output from write_fhx() to include a space between the data block and years. (@chguiterman, #162)
Add flexibility to read_fhx() to handle extra space at bottom of file. This is commonly present in some non-standardized FHX files available on the IMPD. (@chguiterman, #162, #164)
Add DOI badge to the README. (@chguiterman, #162)
Add a new example to get_year() documentation to aid users subsetting an fhx object based on before/after a particular cut-off year. (@chguiterman, #173)
Update count_event_position() to accommodate non-standard positions in rec_type. (@chguiterman, #161)
Add “spring dormant” to the set of rec_types. This is a non-standard descriptor for fire scar seasonality. This aids with cases where "D" dormant scars as the unknown year of a dormant scar (in the Northern Hemisphere). (@chguiterman, #161)
Update README creation to Rmarkdown (.Rmd) file with usethis::use_readme_rmd(). (@chguiterman, #161)
Fix typo in citation file. (@chguiterman, #161)
Add general unit tests to increase test coverage.
Changes in this release:
Added as_fhx() (Issue #120). This takes data frames, tibbles, and lists as input. It assumes they have “year”, “series”, and “rec_type” elements/columns. It returns an fhx object. Additionally, the fhx() constructor now also uses type casting for input (Issue #150) - so no more worrying if your rec_types are factors or character vectors. These changes are a big improvement for anyone working in the tidyverse or anyone working with thier own non-FHX fire-history file formats.
Extensive improvement to documentation (e.g. Issue #145). This includes new “See Also” sections (so users can find cool functions), fixes for spelling errors, and clarifications to dyslexic prose.
write_fhx() will now throw a warning if users try to write an fhx object that has record types violating the FHX2 file convention (Issue #149). I still strongly recommend using write.csv(...) on fhx objects and as.fhx(read.csv(...)) for IO with experimental fhx data.
Removed deprecated run_sea(). Be sure to use sea() now.
Removed deprecated get_ggplot(). Please use plot_demograph() now.
Removed broken site_stats() function (Issue #138). Please use intervals() and print() to get the same statistics.
composite() now returns an empty fhx object if no composite-worthy events are found (Issue #131) or if there are no fire-events (Issue #155). Much better than throwing an obtuse error, which is what we used to do.
Updated in-package citation information (citation("burnr")). Please cite burnr if you use it in your work!
yearly_recording() now returns a data frame with a numeric “year” column (Issue #154). In past versions, the “year” column contained factors.
Added unit tests for basic plotting function options. We now have test coverage for more than 75% of our code.
Minor internal code cleanup (Issue #130, Issue #88, Issue #133, Issue #136, Issue #88, Issue #146) and code linting.
Changes in this minor release:
read_fhx() is more tolerant of extra whitespace in series names. Thanks to Chris Guiterman for adding this.
Fixed documentation examples not rendering for sea() (Issue #125).
Minor argument clarification in composite() documentation (Issue #127).
Changes in this patch:
Updates to documentation and README.
Quick fixes to SEA unit tests, forcing them to use R 3.5.0 RNG as a temporary fix to issue #124.
Changes in this minor release:
Change to sea objects, added event_years.
Added function percent_scarred() for calculating time series of number trees recording, scarred, and the percentage.
Update to sample_depth() to account for single-series fhx objects
Several updates to documentation, including citation.
Changes in this patch:
Added warnings and errors to catch bad input on seas() (#112).
Added warning if using series_mean_interval() on fhx object with multiple series (#113).
Fixed bad numbers in print() for SEAs objects (#111).
Fixed typos in warning messages (#114).
Changes in this patch:
intervals objects’ KS test for distribution test is now two sided (Issue #108).
Printed intervals objects now report the Weibull median, regardless of fit chosen (Issue #109).
Changes in this minor release:
Renamed run_sea() to sea() and enhanced speed, and changed the time_span parameter to event_range, key argument is now event, and years_before and years_after are now nbefore and nafter, respectively. The function now returns a proper sea object with special plot() and print() methods. The plot() generic method uses the new plot_sealags(). You can still use the old run_sea(), but it is deprecated and will be removed in future releases. We’re happy with the general interfaces to sea() but its internal structure may change over the next few minor releases, so beware.
Added intervals() to handle interval analysis. These are a whole new object. mean_interval() has been renamed series_mean_interval() to avoid confusion. We also have a generic plot() function that uses plot_intervals_dist(). It plots a little histogram with a rug along the x-axis.
Added pgm_pdsi data to run with sea. Load it with data("pgm_pdsi").
Added make_rec_type(). I hope this will make it easier to create rec_type factors used by fhx().
Added event_count_position(), which gives the number of events found in different positions or seasons.
Added year_range() to give (min, max) years for an fhx object.
Updated composite() to include a filter for number of events. This necessitates fixes in other functions (e.g. plot_demograph()).
Added summary() function for fhx and intervals objects.
count_fire() renamed count_scar()
Added quantile() function for intervals.
Changes in this patch:
Fixed additional bug in read_fhx() caused by “FHX2 format” in FHX file header.
Fixed bug #71, causing read_fhx() to fail when FHX has empty lastline.
Added text option to read_fhx().
Corrected error message for series arg in fhx(). Fixes #73.
Fixed write_fhx() reversing order of series names in output file.
Changes in this patch:
Added option to sort series by their last (or most recent) year with sort().
Updated run_sea() to better replicate EVENT Fortran program. Addresses #58. Added detailed description and an example page with graphics. Also fixed issue with percentage CIs in the departure table. Mind you, run_sea() is still super experimental.
Several minor changes to documentation and meta data.
Changes in this minor release:
Added a vignette!
get_ggplot() is being depreciated. Please use plot_demograph().
Changed design of fhx objects. They are now data.frame and not lists (nobody used the ‘meta’, list member anyways). This means that if x is your fhx object and you previously used x$rings, you can now just use x.
Added function is.fhx().
Updated pgm example data.
composite() now returns an fhx object.
Added function get_event_years() to find fire and fire injury years in fhx objects. This is especially helpful for composites.
concatenate() and combine() were reverted back to ‘+’ at the request of beloved users. This means code like lgr2 + pgm will combine these objects together into a single fhx object!
Added examples to all function documentation.
resolve_duplicates() is now hidden (i.e. burnr:::check_duplicates()).
Added function sample_depth() to get sample depth.
Added run_sea function to begin to address issue #58. Has pretty table output. This is still very experimental.
Bug fix in series_stats to maintain series IDs for those with no features
Changed composite() behavior. Now more similar to FHAES. Can choose to count injuries as “fire events” with the injury_event argument.
Added count_recording() function.
Fixed minor bugs in series_stats.
Removed sorting by default on read_fhx() and ‘+.fhx’
Fixed sort()s non-responsive decreasing argument.
Fixed poor sorting function for FHX objects.
Added several small utility functions: get_year(), get_series(), series_names(), delete().
Improved documentation!
Adding series_stats function, closes issue #44.
Changes in this patch:
plot() and get_ggplot() should no longer allow the user to plot facets and composite rugs in the same plot. This closes bug #25.
Ring types (the rec_type column in x$rings) has “.” in all levels replaced with a "_". For example, ‘dormant.fs’ is now ‘dormant_fs.’ Closes #45.
In get_ggplot() the legend argument is now plot_legend to avoid clash with common functions.
In fhx$rings and all functions, type argument is now rec_type to avoid clash with type function. rec_type is short for “record_type”.
Can now read FHX files with a single series (closes bug #43).
lgr2 and pgm are available fire-history datasets when you load this package. The metadata for each of these sets is lgr2_meta and pgm_meta. Load the data with data().
First github release
Added NEWS.md file. Existing users should check back here for changes.
rug.filter function is renamed composite.
ggplot.fhx is now get_ggplot. This is more descriptive. The following arguments for this function have been changed as a part of this fix: plot.rug is now composite_rug, filter.min is now filter_min, filter.prop is now filter_prop, event.size is now event_size, rugbuffer.size is now rugbuffer_size, rugdivide.pos is now rugdivide_pos. spp argument is renamed to color_group. sppid is now color_id. cluster is now facet_group. clusterid is now cluster_id.
read.fhx function is renamed read_fhx.
write.fhx function is renamed write_fhx.
order.fhx is now sort (see method sort.fhx). Also, added a boolean argument decreasing, which defaults to FALSE. Previously this has been TRUE..
+ with fhx objects, (+.fhx) is now concatenate. So, a + b should now be concatenate(a, b).
Fixed bug that disappeared a series which started or ended with fire scars.