fit_dsmm()create_sequence()get_kernel()parametric_dsmm()nonparametric_dsmm() (@Bisaloo, 2)CODE_OF_CONDUCT.md file can be
found in the .github folder.get_kernel() has small changes in the description,
regarding the reduction of dimensions when selecting a specific argument
of u, v, l or t.The Github workflow R-hub v2 is added in
.github/workflows. This ensures that after every github
update of the online dsmmR repository, the CRAN checks are
being made for multiple platforms (linux, macos, windows). This can be
run manually through the
command rhub::rhub_check(branch = 'master').
Another Github workflow codecov was added in
.github/workflows. This ensures that the dsmmR
package is mostly covered by the automated tests in place.
valid_p_dist(), valid_fdist_nonparametric(),
valid_fdist_parametric() and added a cases for the f
non-drifting case in get_fdist_parametric(). This was
causing some errors to appear when trying to print an estimated fitted
model when f was not drifting (Model 2).paper.bib, DOIs were added
in the correct formatting. This was a small fix done for the JOSS
publication.fit_dsmm() gains a new attribute
multi_estimation, which enables the estimation of a
drifting semi-Markov model using multiple sequences. There are two
possible options: avg_model and count_sum.
avg_model averages the q_i received from
multiple sequencescount_sum adds the counts of the states for each
sequence (of equal size) and then computes the q_i.simulate.dsmm() gains a new attribute,
max_seq_length, which is renamed from old attribute
seq_length for clarity.Depends section. Now we impose the
requirement for R >= 3.5.0, in order to make proper use of the
isTRUE() and isFALSE() functions in the
is_logical() function defined in utils.R.
These functions will remain for their clarity.fit_dsmm() and simulate.dsmm() functions
now better explain the difference between a sequence of states and the
embedded Markov chain. Notably, it was specified that sequences are
input in fit_dsmm(), specified with the argument
sequence. In simulate.dsmm(), such a sequence
is the resulting output. The embedded Markov chain can be seen as part
of the output from fit_dsmm(), named emc.
Furthermore, the function base::rle() is mentioned for
clarity.README and DESCRIPTION files
with an acknowledgement section.Added a NEWS.md file to track changes to the
package.
Now the fit_dsmm() function has a default value for
the states attribute, being the sorted unique values of the
sequence character vector attribute.
simulate.dsmm() sometimes did not
function as expected when nsim = 1.
Now it is possible to specify nsim = 0, so that the
simulated sequence will only include the initial state and its
corresponding sojourn time, e.g. “a”, “a”, “a”.
By giving nsim = 1 , a single simulation will be made
from the drifting semi-Markov kernel, returning for example “a”, “a”,
“a”, “c”.
Updated the documentation for simulate.dsmm(), with
accordance to the changes made.
Updated the README file.
Added high-level documentation of the package.
Added installation instructions with access to the development version of the package through github.
Updated the documentation for dsmmR-package.
Added a “Community Guidelines” section, so that users can report errors or mistakes and contribute directly to the software through the newly-established open-source github page at https://github.com/Mavrogiannis-Ioannis/dsmmR.
Added a “Notes” section, specifying that automated tests are in place in order to aid the user with any false input made and, furthermore, to ensure that the functions used return the expected output.