the case of 3 < n < 8 has been removed from ctbi.outlier
The companion paper has been published with final DOI and added to
ctbi
ctbi 2.0.4 (2022/11/18)
Fixed a minor mistake concerning the way residuals are handled for
limited range of possible values (ex. [0,+Inf[ instead of
]-Inf,+Inf[)
The description of the functions has been updated
A temporary DOI has been attached until the review of the original
manuscript is finished
ctbi 2.0.3 (2022/07/27)
summary.bin has been added to the output of ctbi, and it condenses
the information about the size of a bin (bin.size), the minimum number
of points for a bin to be accepted (bin.size.min.accepted) and the
Stacked Cycle Index (SCI)
the mean of mean cycle has been subtracted and added to the
long.term
examples have been updated
ctbi 2.0.2 (2022/07/25)
summary.outlier has been added to the output of ctbi, and gives
information about coeff.outlier (coefficients A,B and C), the
lower/upper outlier threshold, the number of points used, the value of
m.star.
a minor mistake has been fixed for the residuals
the outliers are now present in the column residuals to compare them
to the lower/upper threshold
ctbi 2.0.1 (2022/07/22)
ctbi.outliers has been changed with ctbi.outlier. This function now
only takes an input vector and can handle any univariate datasets (no
necessarily based on residuals).
The Logbox has been updated.
g_A(m.star)=0.2294exp(2.9416m.star-0.0512m.star2-0.0684m.star3)
and
g_B(m.star)=1.0585+15.6960m.star-17.3618m.star2+28.3511m.star3-11.4726m.star^4
will respectively calculate the A and B coefficients with m.star =
(q(0.875)-q(0.625))/(q(0.75)-q(0.25))-0.6165 for right-skewed
distributions, bounded by [0,2]. The C coefficients is fixed to 50
(Pearson Family).
A “gaussian” method for Logbox has been added, with coeff.outlier =
c(0.08,2,33)
for n < 9, the MAD will be used instead the boxplot rule, with a
multiplying factor of 10 (instead of usually 3) to limit the number of
false positive to 0.5%.
ctbi 2.0.0 (2022/06/10)
The LogBox method has changed from alpha = 1+klog(n) to alpha =
Alog(n)+B+C/n
k.outliers has been replaced by coeffs.outlier = c(A,B,C).
SCI.min = Inf has been replaced by SCI.min = NA when no values are
imputed
ctbi 1.0.1 (2022/03/08)
F or T have been replaced with FALSE or TRUE
Example 2 has been added to the help of ctbi
Value has been added to the help of ctbi.plot and hidd.seq
ctbi 1.0.0 (2022/03/01)
ORCID of Francois Ritter has been updated.
n.bin.min <- floor(bin.size*(1-bin.max.f.NA)) has been replaced
with n.bin.min <- ceiling(bin.size*(1-bin.max.f.NA)). This means that
for bin.max.f.NA = 0.2 (bins with at least 80% of data are accepted) and
bin.size = 12 (e.g., 12 months of data per bin), a bin with 9 months of
data will be rejected.
n.bin.min (minimum number of data points for a bin to be accepted)
has been added as an output of ctbi in list.main.
The starting and ending boundaries of the long.term interpolation
were treated as NA values. This has been removed.
The numeric class of data0[,y] has been forced to avoid problems
with integers.
Fixed a minor problem with the error message concerning
k.outliers