sg_apply(dt_prep_sets)
: add extra"_row<row-number>"
string toprep_label
column ofdt_prep_sets
input when combining with Savitzky-Golay parameter sets. This modification makes sure that cartesian products of existing preprocessing sets supplied as input are correctly formed with the repeated Savitzky-Golay plans.
sg_apply(dt_prep_sets)
: There was an intermediary list-column calledprep_params_in
, that was outputted whendt_prep_sets
was supplied as argument. This column is now omitted in the output data.table with the updated preprocessing sets, so that the objects produced can be easily row-bound with later objects downstream, or combined with the input object whenappend_rows = TRUE
.
- failed to commit changes. Keeping entry for reproducibility. The intended fix is is in version 0.3.4 (see above).
snv_apply(X)
: addprep_params
as list-column with a single-row data.table (snv = NA
) to the output whenX
is provided. This makes binding outputs to other (pre)processed collections of spectra possible without further intervention; also,append_rows = TRUE
will work with other methods, when output ofsnv_apply()
is used as input of other*_apply()
functions.
sg_apply()
: allow joins of Savitzky-Golay plans and preprocessing labels, and then also prepared Savitzky-Golay plans with inputteddt_prep_sets
, when there is duplicatedi
s. This is the case whendt_prep_sets
input has already multiple rows (multiple collections) of spectra. Now, the desired duplicate joins are explicitly allowed by settingallow.cartesian = TRUE
for respective data.table joins inside thesg_make_dt_prep()
helper.
- Implement
colmean_group_apply()
and group label constructorids_apply.R()
. This is to apply column means to to spectral collections, each by a group label.
- patch
sg_apply()
so that the extra"-snv"
that got accidentally added to bothprep_set
andprep_label
is not there anymore.
- patch
sg_apply()
, so that it can be run after e.g.snv_apply()
(viadt_prep_sets
input argument).
- added
snv_apply()
to compute the standard normal variate (SNV) of spectral collections (#15). - added
sg_apply()
to process spectral collections with Savitzky-Golay smoothers with different parameter sets (derivative order, window size, polynomial degree).
- Started semantic versioning via {fledge}
Chemometrics and machine learning offer a large set of mathematical tooling to extract and apply chemical and physical knowledge from spectra in automated fashion. For this, spectra are typically preprocessed as part of the workflow. This is mostly to reduce light scattering and other optical artefacts.
The goal of {specprepper} is not only to wrap different signal processing methods and make them more accessible, but also to offer some of the exisiting algorithms with faster code implementations. It features a recipe-like interface, which also makes it possible to chain different methods in sequence.