expirest 0.1.7

• The check_ancova() function was modified to favor the different intercepts /
  different slopes (dids) model when ANCOVA suggests that the common
  intercepts / different slopes (cids) model is appropriate, as the cids model
  is not practically relevant.
• References to the corresponding plotting functions was added to the help
  sections of the expirest_osle() and expirest_wisle() function.
• To the help sections of all internal functions the @nord tag was added.
• Remove expect_length() tests that were used for ggplot graph labels. A
  planned update of ggplot2 would affect these tests because ggplot2 will
  not populate the plot$labels field before plot building anymore. For this
  reason, the test assumptions would be violated. The tests are thus removed
  to prevent errors.

expirest 0.1.6

This release comes with the following changes:

• So far, exclusively the result from fitting individual models to each batch
  were reported in case of the different intercepts / different slopes (dids).
  model. With the new version, the results of the dids model obtained with
  pooled mean square error (dids.pmse) will be shown in addition. The results
  of the dids and the dids.pmse are reported generally in summaries.
• A new parameter has been added to the two plot functions allowing to select
  the model to be displayed. The show_grouping parameter is not relevant any
  more and is thus marked as deprecated.
• So far, the shelf life assessment failed when the results of a single batch
  were used. The new version allows to perform the assessment with a single
  batch, and the corresponding results are reported as dids model (while the
  model type will be n.a.).
• The function ggplot2::coord_cartesian() is used to prevent the loss of
  information displayed on graphs when the intervals have values going beyond
  the edges of the plot.
• In case of a specification with two sides, the shelf life assessment had to
  be done for each side one at a time. In the new version, the ordinary shelf
  life estimation will be performed on both sides, and the shortest shelf life
  result will be reported (together with the information about the side). For
  the what-if shelf life analysis, this does not apply because for this kind of
  assessment the critical side is known.
• The requirements on rl has been relaxed by taking into account the number
  of significant digits specified for the specification limit (sl_sf). For
  example, if ivl_side = "upper", sl = 0.5 and sl_sf = 1, rl values between
  0.50 and 0.54 are accepted because the corresponding rounded rl values (to
  one significant digit) all are 0.5 and thus not greater than 0.5.
• An error in the calculation of sl and rl in case of sf_option = "loose" was
  corrected. The correction make sure that 5 is subtracted at the (sl_sf + 1)th
  or (rl_sf + 1)th figure if ivl_side is "lower" or that 4 is added at the
  (sl_sf + 1)th or (rl_sf + 1)th figure if ivl_side is "upper".
• Cyclomatic complexity was reduced as far as possible, i.e. the complexity of
  functions was reduced in favor of more (internal) functions.
• Examples have been slimmed down and the execution of some of the example
  code is prevented to reduce the execution time of examples.

expirest 0.1.5

• Expired and (possibly) invalid URLs were removed.
• Due to a misconception of object type tests the same value was reported for
  each rl value if the cics model was the appropriate model although the
  correct values were calculated. Now, the correct values (i.e. Delta, WCSL
  and Shelf.Life) are reported.
• The output of the summary() function has been harmonised between cics models
  on the one hand and dics / dids models on the other.
• The deprecated ggplot2 function aes_string() has been replaced by aes().
  Variables containing column names as a character vector (var_name) are
  now called via .data[[var_name]]. In addition, size has been replaced by
  linewidth when used in the geom_line() function. In addition, visible binding
  for global variables was added.

expirest 0.1.3

• The graphical output of the dids model was not appropriate because the
  prediction was based on the whole model with interaction term (group x time,
  where group is a categorical variable and time is a continuous variable).
  Now, the prediction is based on individual models fitted to the different
  levels of group.
• Before, a decimal point was shown in case of numbers without decimal digits.
• Comparisons of type class() == "string" were replaced by inherits() calls.

First CRAN release

expirest 0.1.2

• CRAN release
• Examples that were wrapped by \dontrun{} although they are executable in
  < 5 sec are now unwrapped.
• Errors from examples that deliberately produce errors are caught by tryCatch()
  and are no longer wrapped by \dontrun{}.