| Type: | Package |
| Title: | High-Dimensional Analysis of Variance |
| Version: | 0.8.6 |
| Date: | 2026-11-20 |
| Description: | Functions and datasets to support Smilde, Marini, Westerhuis and Liland (2025, ISBN: 978-1-394-21121-0) "Analysis of Variance for High-Dimensional Data - Applications in Life, Food and Chemical Sciences". This implements and imports a collection of methods for HD-ANOVA data analysis with common interfaces, result- and plotting functions, multiple real data sets and four vignettes covering a range different applications. |
| Depends: | R (≥ 3.5.0) |
| Imports: | car, lme4, MASS, mixlm (≥ 1.4.3), pls, pracma, progress, Rcpp, reformulas, RSpectra |
| LinkingTo: | Rcpp, RcppArmadillo |
| Suggests: | knitr, testthat (≥ 3.0.0), vegan |
| Config/testthat/edition: | 3 |
| License: | GPL-2 | GPL-3 [expanded from: GPL (≥ 2)] |
| URL: | https://khliland.github.io/HDANOVA/, https://github.com/khliland/HDANOVA/ |
| BugReports: | https://github.com/khliland/HDANOVA/issues/ |
| Encoding: | UTF-8 |
| LazyData: | true |
| VignetteBuilder: | knitr |
| Config/roxygen2/version: | 8.1.0 |
| NeedsCompilation: | yes |
| Packaged: | 2026-11-20 15:50:12 UTC; kristian |
| Author: | Kristian Hovde Liland
|
| Maintainer: | Kristian Hovde Liland <kristian.liland@nmbu.no> |
| Repository: | CRAN |
| Date/Publication: | 2026-11-20 17:30:12 UTC |
High-Dimensional Analysis of Variance
Description
Included methods:
ASCA+ - Analysis of Variance Simultaneous Component Analysis
APCA+ - ANOVA Principal Component Analysis
APLS+ - ANOVA Partial Least Squares
LiMM-PCA - Linear Mixed Model PCA
MSCA - Multilevel Simultaneous Component Analysis
PC-ANOVA - Principal Component Analysis of Variance
PRC - Principal Response Curves
PERMANOVA - Permutation Based MANOVA
Author(s)
Maintainer: Kristian Hovde Liland kristian.liland@nmbu.no (ORCID)
Authors:
Kristian Hovde Liland kristian.liland@nmbu.no (ORCID)
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
ANOVA Principal Component Analysis - APCA
Description
APCA function for fitting ANOVA Principal Component Analysis models.
Usage
apca(
formula,
data,
add_error = TRUE,
contrasts = "contr.sum",
permute = FALSE,
perm.type = c("approximate", "exact"),
...
)
Arguments
formula |
Model formula accepting a single response (block) and predictors. |
data |
The data set to analyse. |
add_error |
Add error to LS means (default = TRUE). |
contrasts |
Effect coding: "sum" (default = sum-coding), "weighted", "reference", "treatment". |
permute |
Number of permutations to perform (default = 1000). |
perm.type |
Type of permutation to perform, either "approximate" or "exact" (default = "approximate"). |
... |
Additional parameters for the |
Value
An object of class apca, inheriting from the general asca class.
Further arguments and plots can be found in the asca documentation.
References
Harrington, P.d.B., Vieira, N.E., Espinoza, J., Nien, J.K., Romero, R., and Yergey, A.L. (2005) Analysis of variance–principal component analysis: A soft tool for proteomic discovery. Analytica chimica acta, 544 (1-2), 118–127.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
data(candies)
ap <- apca(assessment ~ candy, data=candies)
scoreplot(ap)
# Numeric effects
candies$num <- eff <- 1:165
mod <- apca(assessment ~ candy + assessor + num, data=candies)
summary(mod)
scoreplot(mod, factor=3, gr.col=rgb(eff/max(eff), 1-eff/max(eff),0), pch.scores="x")
Analysis of Variance Partial Least Squares - APLS
Description
This is a quite general and flexible implementation of APLS.
Usage
apls(
formula,
data,
add_error = TRUE,
contrasts = "contr.sum",
permute = FALSE,
perm.type = c("approximate", "exact"),
...
)
Arguments
formula |
Model formula accepting a single response (block) and predictors. See Details for more information. |
data |
The data set to analyse. |
add_error |
Add error to LS means (default = TRUE). |
contrasts |
Effect coding: "sum" (default = sum-coding), "weighted", "reference", "treatment". |
permute |
Number of permutations to perform (default = 1000). |
perm.type |
Type of permutation to perform, either "approximate" or "exact" (default = "approximate"). |
... |
Additional arguments to |
Details
APLS is a method which decomposes a multivariate response according to one or more design
variables. ANOVA is used to split variation into contributions from factors, and PLS is performed
on the corresponding least squares estimates, i.e., Y = X1 B1 + X2 B2 + ... + E = T1 P1' + T2 P2' + ... + E.
For balanced designs, the PLS components are equivalent to PCA components, i.e., APLS and APCA are equivalent.
This version of APLS encompasses variants of LiMM-PLS, generalized APLS and covariates APLS.
The formula interface is extended with the function r() to indicate random effects and comb() to indicate effects that should be combined. See Examples for use cases.
Value
An apls object containing loadings, scores, explained variances, etc. The object has
associated plotting (asca_plots) and result (asca_results) functions.
References
Smilde, A., Jansen, J., Hoefsloot, H., Lamers,R., Van Der Greef, J., and Timmerman, M.(2005). ANOVA-Simultaneous Component Analysis (ASCA): A new tool for analyzing designed metabolomics data. Bioinformatics, 21(13), 3043–3048.
Liland, K.H., Smilde, A., Marini, F., and Næs,T. (2018). Confidence ellipsoids for ASCA models based on multivariate regression theory. Journal of Chemometrics, 32(e2990), 1–13.
Martin, M. and Govaerts, B. (2020). LiMM-PCA: Combining ASCA+ and linear mixed models to analyse high-dimensional designed data. Journal of Chemometrics, 34(6), e3232.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# Load candies data
data(candies)
# Basic APLS model with two factors
mod <- apls(assessment ~ candy + assessor, data=candies)
print(mod)
# APLS model with interaction
mod <- apls(assessment ~ candy * assessor, data=candies)
print(mod)
# Result plotting for first factor
loadingplot(mod, scatter=TRUE, labels="names")
scoreplot(mod)
# No backprojection
scoreplot(mod, projections=FALSE)
# Spider plot
scoreplot(mod, spider=TRUE)
# APLS model with compressed response using 5 principal components
mod.pca <- apls(assessment ~ candy + assessor, data=candies, pca.in=5)
# Mixed Model APLS, random assessor
mod.mix <- apls(assessment ~ candy + r(assessor), data=candies)
scoreplot(mod.mix)
# Mixed Model APLS, REML estimation
mod.mix <- apls(assessment ~ candy + r(assessor), data=candies, REML=TRUE)
scoreplot(mod.mix)
# Load Caldana data
data(caldana)
# Combining effects in APLS
mod.comb <- apls(compounds ~ time + comb(light + time:light), data=caldana)
summary(mod.comb)
timeplot(mod.comb, factor="light", time="time", comb=2)
# Permutation testing
mod.perm <- apls(assessment ~ candy * assessor, data=candies, permute=TRUE)
summary(mod.perm)
Analysis of Variance Simultaneous Component Analysis - ASCA
Description
This is a quite general and flexible implementation of ASCA.
Usage
asca(
formula,
data,
contrasts = "contr.sum",
permute = FALSE,
perm.type = c("approximate", "exact"),
...
)
Arguments
formula |
Model formula accepting a single response (block) and predictors. See Details for more information. |
data |
The data set to analyse. |
contrasts |
Effect coding: "sum" (default = sum-coding), "weighted", "reference", "treatment". |
permute |
Number of permutations to perform (default = 1000). |
perm.type |
Type of permutation to perform, either "approximate" or "exact" (default = "approximate"). |
... |
Additional arguments to |
Details
ASCA is a method which decomposes a multivariate response according to one or more design
variables. ANOVA is used to split variation into contributions from factors, and PCA is performed
on the corresponding least squares estimates, i.e., Y = X1 B1 + X2 B2 + ... + E = T1 P1' + T2 P2' + ... + E.
This version of ASCA encompasses variants of LiMM-PCA, generalized ASCA and covariates ASCA. It includes
confidence ellipsoids for the balanced crossed-effect ASCA.
The formula interface is extended with the function r() to indicate random effects and comb() to indicate effects that should be combined. See Examples for use cases.
Value
An asca object containing loadings, scores, explained variances, etc. The object has
associated plotting (asca_plots) and result (asca_results) functions.
References
Smilde, A., Jansen, J., Hoefsloot, H., Lamers,R., Van Der Greef, J., and Timmerman, M.(2005). ANOVA-Simultaneous Component Analysis (ASCA): A new tool for analyzing designed metabolomics data. Bioinformatics, 21(13), 3043–3048.
Liland, K.H., Smilde, A., Marini, F., and Næs,T. (2018). Confidence ellipsoids for ASCA models based on multivariate regression theory. Journal of Chemometrics, 32(e2990), 1–13.
Martin, M. and Govaerts, B. (2020). LiMM-PCA: Combining ASCA+ and linear mixed models to analyse high-dimensional designed data. Journal of Chemometrics, 34(6), e3232.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# Load candies data
data(candies)
# Basic ASCA model with two factors
mod <- asca(assessment ~ candy + assessor, data=candies)
print(mod)
# ASCA model with interaction
mod <- asca(assessment ~ candy * assessor, data=candies)
print(mod)
# Result plotting for first factor
loadingplot(mod, scatter=TRUE, labels="names")
scoreplot(mod)
# No backprojection
scoreplot(mod, projections=FALSE)
# Spider plot
scoreplot(mod, spider=TRUE)
# ASCA model with compressed response using 5 principal components
mod.pca <- asca(assessment ~ candy + assessor, data=candies, pca.in=5)
# ASCA model with design-relevant compressed response
# using 5 partial least squares components
mod.pca <- asca(assessment ~ candy + assessor, data=candies, pls.in=5)
# Mixed Model ASCA, random assessor
mod.mix <- asca(assessment ~ candy + r(assessor), data=candies)
scoreplot(mod.mix)
# Mixed Model ASCA, REML estimation
mod.mix <- asca(assessment ~ candy + r(assessor), data=candies, REML=TRUE)
scoreplot(mod.mix)
# Load Caldana data
data(caldana)
# Combining effects in ASCA
mod.comb <- asca(compounds ~ time + comb(light + time:light), data=caldana)
summary(mod.comb)
timeplot(mod.comb, factor="light", time="time", comb=2)
# Permutation testing
mod.perm <- asca(assessment ~ candy * assessor, data=candies, permute=TRUE)
summary(mod.perm)
# Unbalanced data: compare native vs SS-type-aligned effects
drop_idx <- which(candies$candy == levels(candies$candy)[1] &
candies$assessor == levels(candies$assessor)[1])
candies.unbal <- candies[-drop_idx[seq_len(min(2, length(drop_idx)))], ]
mod.native <- asca(assessment ~ candy * assessor, data = candies.unbal)
mod.sstype <- asca(assessment ~ candy * assessor, data = candies.unbal,
respect_SStype = TRUE)
par.old <- par(mfrow = c(2,1), mar = c(4,4,2,1))
scoreplot(mod.native,
main = "Top: native (regression-based LS)")
scoreplot(mod.sstype,
main = "Bottom: respect_SStype = TRUE")
par(par.old)
perm.native <- permutation(mod.native, permute = 1000, respect_SStype = FALSE)
perm.sstype <- permutation(mod.sstype, permute = 1000, respect_SStype = TRUE)
summary(perm.native)
summary(perm.sstype)
ASCA Plot Methods
Description
Various plotting procedures for asca objects.
Usage
## S3 method for class 'asca'
loadingplot(object, factor = 1, comps = 1:2, ...)
## S3 method for class 'asca'
scoreplot(
object,
factor = 1,
comps = 1:2,
within_level = "all",
pch.scores = 19,
pch.projections = 1,
gr.col = NULL,
projections = TRUE,
spider = FALSE,
ellipsoids,
confidence,
xlim,
ylim,
xlab,
ylab,
legendpos,
...
)
permutationplot(object, factor = 1, xlim, xlab = "SSQ", main, ...)
rotationplot(object, factor = 1, xlim, xlab = "SSQ", main, ...)
Arguments
object |
|
factor |
|
comps |
|
... |
additional arguments to underlying methods. |
within_level |
MSCA parameter for chosing plot level (default = "all"). |
pch.scores |
|
pch.projections |
|
gr.col |
|
projections |
Include backprojections in score plot (default = TRUE). |
spider |
Draw lines between group centers and backprojections (default = FALSE). |
ellipsoids |
|
confidence |
|
xlim |
|
ylim |
|
xlab |
|
ylab |
|
legendpos |
|
main |
Plot title. |
Details
Usage of the functions are shown using generics in the examples in asca.
Plot routines are available as
scoreplot.asca and loadingplot.asca.
Value
The plotting routines have no return.
References
Smilde, A., Jansen, J., Hoefsloot, H., Lamers,R., Van Der Greef, J., and Timmerman, M.(2005). ANOVA-Simultaneous Component Analysis (ASCA): A new tool for analyzing designed metabolomics data. Bioinformatics, 21(13), 3043–3048.
Liland, K.H., Smilde, A., Marini, F., and Næs,T. (2018). Confidence ellipsoids for ASCA models based on multivariate regression theory. Journal of Chemometrics, 32(e2990), 1–13.
Martin, M. and Govaerts, B. (2020). LiMM-PCA: Combining ASCA+ and linear mixed models to analyse high-dimensional designed data. Journal of Chemometrics, 34(6), e3232.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# For end-to-end examples using scoreplot()/loadingplot(), see
# ?asca, ?apca, ?apls, ?limmpca and ?msca.
ASCA Result Methods
Description
Standard result computation and extraction functions for ASCA (asca).
Usage
## S3 method for class 'hdanova'
print(x, ...)
## S3 method for class 'hdanova'
summary(object, extended = TRUE, df = FALSE, ...)
## S3 method for class 'summary.hdanova'
print(x, digits = 2, ...)
## S3 method for class 'asca'
loadings(object, factor = 1, ...)
## S3 method for class 'asca'
scores(object, factor = 1, ...)
projections(object, ...)
## S3 method for class 'asca'
projections(object, factor = 1, ...)
Arguments
x |
|
... |
additional arguments to underlying methods. |
object |
|
extended |
Extended output in summary (default = TRUE). |
df |
Show degrees of freedom in summary (default = FALSE). |
digits |
|
factor |
|
Details
Usage of the functions are shown using generics in the examples in asca.
Explained variances are available (block-wise and global) through blockexpl and print.rosaexpl.
Object printing and summary are available through:
print.asca and summary.asca.
Scores and loadings have their own extensions of scores() and loadings() through
scores.asca and loadings.asca. Special to ASCA is that scores are on a
factor level basis, while back-projected samples have their own function in projections.asca.
Value
Returns depend on method used, e.g. projections.asca returns projected samples,
scores.asca return scores, while print and summary methods return the object invisibly.
References
Smilde, A., Jansen, J., Hoefsloot, H., Lamers,R., Van Der Greef, J., and Timmerman, M.(2005). ANOVA-Simultaneous Component Analysis (ASCA): A new tool for analyzing designed metabolomics data. Bioinformatics, 21(13), 3043–3048.
Liland, K.H., Smilde, A., Marini, F., and Næs,T. (2018). Confidence ellipsoids for ASCA models based on multivariate regression theory. Journal of Chemometrics, 32(e2990), 1–13.
Martin, M. and Govaerts, B. (2020). LiMM-PCA: Combining ASCA+ and linear mixed models to analyse high-dimensional designed data. Journal of Chemometrics, 34(6), e3232.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# For end-to-end examples using summary(), scores(), loadings(), and
# projections(), see ?asca and ?apca.
Biplot for ASCA models
Description
Biplot for ASCA models
Usage
## S3 method for class 'asca'
biplot(
x,
factor = 1,
comps = 1:2,
xlim = NULL,
ylim = NULL,
col = "darkgray",
expand = 1,
labels,
legendpos,
...
)
Arguments
x |
|
factor |
Factor number or name. |
comps |
|
xlim |
|
ylim |
|
col |
|
expand |
|
labels |
optional. If |
legendpos |
|
... |
Additional arguments to |
Value
No return, only a plot.
Examples
# Load candies data
data(candies)
# Basic ASCA model with two factors and interaction
mod <- asca(assessment ~ candy * assessor, data=candies)
# Biplot
biplot(mod)
# Biplot with named loadings
biplot(mod, labels="names")
Block-wise indexable data.frame
Description
This is a convenience function for making data.frames that are easily
indexed on a block-wise basis.
Usage
block.data.frame(X, block_inds = NULL, to.matrix = TRUE)
Arguments
X |
Either a single |
block_inds |
Named |
to.matrix |
|
Value
A data.frame which can be indexed block-wise.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# Random data
M <- matrix(rnorm(200), nrow = 10)
# .. with dimnames
dimnames(M) <- list(LETTERS[1:10], as.character(1:20))
# A named list for indexing
inds <- list(B1 = 1:10, B2 = 11:20)
X <- block.data.frame(M, inds)
str(X)
Arabidopsis thaliana growth experiment
Description
A dataset containing 67 metabolites from plants grown under different light and temperature conditions. This subset of the data contains only the light effect and time effect for limited conditions, while the full data also contains gene expressions.
Usage
data(caldana)
Format
A data.frame having 140 rows and 3 variables:
- light
Light levels
- time
Time of measurement
- compound
Metabolic compounds
References
Caldana C, Degenkolbe T, Cuadros-Inostroza A, Klie S, Sulpice R, Leisse A, et al. High-density kinetic analysis of the metabolomic and transcriptomic response of Arabidopsis to eight environmental conditions. Plant J. 2011;67(5):869-884.
Sensory assessment of candies.
Description
A dataset containing 9 sensory attributes for 5 candies assessed by 11 trained assessors.
Usage
data(candies)
Format
A data.frame having 165 rows and 3 variables:
- assessment
Matrix of sensory attributes
- assessor
Factor of assessors
- candy
Factor of candies
References
Luciano G, Næs T. Interpreting sensory data by combining principal component analysis and analysis of variance. Food Qual Prefer. 2009;20(3):167-175.
Dummy-coding of a single vector
Description
Flexible dummy-coding allowing for all R's built-in types of contrasts and optional dropping of a factor level to reduce rank defficiency probability.
Usage
dummycode(Y, contrast = "contr.sum", drop = TRUE)
Arguments
Y |
|
contrast |
Contrast type, default = "contr.sum". |
drop |
|
Value
matrix made by dummy-coding the input vector.
Examples
vec <- c("a","a","b","b","c","c")
dummycode(vec)
Extracting the Extended Model Frame from a Formula or Fit
Description
This function attempts to apply model.frame and extend the
result with columns of interactions.
Usage
extended.model.frame(formula, data, ..., sep = ".")
Arguments
formula |
a model formula or terms object or an R object. |
data |
a data.frame, list or environment (see |
... |
further arguments to pass to |
sep |
separator in contraction of names for interactions (default = "."). |
Value
A data.frame that includes everything a model.frame
does plus interaction terms.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
dat <- data.frame(Y = c(1,2,3,4,5,6),
X = factor(LETTERS[c(1,1,2,2,3,3)]),
W = factor(letters[c(1,2,1,2,1,2)]))
extended.model.frame(Y ~ X*W, dat)
Extract estimates for a given factor combination
Description
Extracts and sums the LS estimates for a given factor combination
from an object of class hdanova. If add_residuals is TRUE,
the residuals are added to the LS estimates. If substract is TRUE,
the returned matrix is the data with chosen estimates subtracted.
Usage
extract_estimates(object, factors, subtract = FALSE, add_residuals = FALSE)
Arguments
object |
|
factors |
|
subtract |
|
add_residuals |
|
Value
A matrix of the extracted estimates.
See Also
Model fitting and related outputs: hdanova,
asca_results and predict.hdanova.
Examples
# Load candies data
data(candies)
# Basic HDANOVA model with two factors and interaction
mod <- hdanova(assessment ~ candy * assessor, data=candies)
# Extract estimates for the interaction
inter <- extract_estimates(mod, c("candy:assessor"))
# Visualize the interaction effect
image(t(inter), main="Interaction effect", xlab="Attribute", ylab="Sample")
High-Dimensional Analysis of Variance
Description
This function provides a high-dimensional analysis of variance (HDANOVA) method
which can be used alone or as part of a larger analysis, e.g., ASCA, APCA, LiMM-PCA, MSCA or PC-ANOVA. It
can be called directly or through the convenience functions asca, apca,
limmpca, msca and pcanova.
Usage
hdanova(
formula,
data,
subset,
weights,
na.action,
family,
scale = FALSE,
add_error = FALSE,
aug_error = "denominator",
pca.in = FALSE,
pls.in = FALSE,
contrasts = "contr.sum",
unrestricted = FALSE,
SStype = "II",
respect_SStype = FALSE,
REML = NULL,
REML_ssq_method = c("exact_refit", "wald", "ls"),
equal_baseline = FALSE,
use_ED = FALSE
)
Arguments
formula |
Model formula accepting a single response (block) and predictors. See Details for more information. |
data |
The data set to analyse. |
subset |
Expression for subsetting the data before modelling. |
weights |
Optional object weights. |
na.action |
How to handle NAs (no action implemented). |
family |
Error distributions and link function for Generalized Linear Models. |
scale |
Scaling of the response matrix. Defaults to |
add_error |
Add error to LS means, e.g., for APCA. |
aug_error |
Augment score matrices in backprojection. Default = "denominator" (of F test), "residual" (force error term), numeric value (alpha-value in LiMM-PCA). |
pca.in |
Compress response before ASCA (number of components). |
pls.in |
Compress response before ASCA using PLS (number of components). |
contrasts |
Effect coding: "contr.sum" (default), "contr.weighted" (not for lme4 models), "contr.reference", "contr.treatment". Can also be specified shorthand without "contr.", e.g., "sum", or as a named vector for different coding per factor. |
unrestricted |
Use unrestricted ANOVA decomposition (default = FALSE). |
SStype |
Type of sum-of-squares: "I" = sequential, "II" (default) = last term, obeying marginality, "III" = last term, not obeying marginality. See Details for more information. |
respect_SStype |
Logical; if |
REML |
Parameter to mixlm: NULL (default) = sum-of-squares, TRUE = REML, FALSE = ML.
If supplied without any |
REML_ssq_method |
Method for REML mixed-model SSQ decomposition:
|
equal_baseline |
Experimental: Set to |
use_ED |
Use "effective dimensions" for score rescaling in LiMM-PCA. |
Details
Scaling of the response matrix can be done by setting the scale parameter. If scale=TRUE,
each column is scaled by its standard deviation (autoscaling). A numeric value can be provided to scale
the columns by specific quantities. If scale is a character string, the first element
is interpreted as a factor name and the second element is interpreted as a factor level, whose samples
the standard deviations are calculated from (reference group scaling).
SStype and respect_SStype (fixed and MoM mixed models):
SStype controls how each effect's contribution is isolated from the others.
Type I (sequential) assigns variance in the order terms appear in the formula;
Type II (default) tests each term against all others that do not contain it, respecting
marginality; Type III tests each term against the full model, ignoring marginality.
For balanced designs the three types give identical results; differences arise in
unbalanced or non-orthogonal designs. Note that in unbalanced designs, SSQ are not n
ecessarily additive, so the sum of effect SSQ may not equal the total SSQ.
By default (respect_SStype = FALSE) the effect matrices in LS are regression
projections — each effect's columns are projected independently from the full model
coefficient matrix. This is consistent with the legacy ASCA workflow. When
respect_SStype = TRUE the LS matrices are re-derived from QR contrasts
that match the chosen SStype, so the Frobenius norm of LS[[effect]] equals
the corresponding ANOVA sum of squares. The regression matrices are still available in
LS_regression. For balanced, fully-crossed designs the two sets are numerically
identical; for unbalanced designs they will differ, particularly for Type II and III.
REML SSQ strategies (mixed models with r() terms):
When REML = TRUE or REML = FALSE, the model is fitted with lme4 and the
classical ANOVA table is replaced by one of three decomposition strategies controlled by
REML_ssq_method:
-
"exact_refit"(default): For each effect a reduced model is re-fitted (REML or ML as specified) and the SSQ is the difference in log-likelihoods scaled to the sum-of-squares metric. Concretely, if\ell_fand\ell_rare the log-likelihoods of the full and reduced models, the statistic is\mathrm{SSQ} = 2(\ell_f - \ell_r).This is the most principled approach but requires one additional model fit per effect, making it slower for large models.
-
"wald": Uses the Wald chi-square statistic from the full model divided by the residual variance to obtain an approximate SSQ. For a fixed effect with coefficient vector\hat{\boldsymbol{\beta}}_jand covariance\mathbf{V}_j = \mathrm{Var}(\hat{\boldsymbol{\beta}}_j),\mathrm{SSQ} \approx \hat{\boldsymbol{\beta}}_j^\top \mathbf{V}_j^{-1} \hat{\boldsymbol{\beta}}_j \cdot \hat{\sigma}^2.No additional model fits are required; fast but less accurate for small samples or near-singular random effects.
-
"ls": Projects the REML/ML coefficient estimates through the fixed-effect design matrix to recover a least-squares-style SSQ,\mathrm{SSQ} = \| \mathbf{X}_j \hat{\boldsymbol{\beta}}_j \|_F^2,where
\mathbf{X}_jcontains the columns of the design matrix for effectj. Computationally cheap and numerically stable but ignores the mixed-model covariance structure; best treated as an approximation.
As in the MoM mixed-model, unbalanced designs and non-orthogonal effects will lead to
non-additive SSQ, so the sum of effect SSQ may not equal the total SSQ. However, as the
REML SSQ are more complex functions of the data than the MoM SSQ, they may differ more
substantially from the regression-based SSQ in LS_regression than in the
fixed-effects case, even for balanced designs.
Permutation statistics vs. fitted-model SSQ:
Permutation testing (permutation) always uses QR-based projection
statistics computed on the fixed-effect design matrix, regardless of REML_ssq_method.
When respect_SStype = FALSE (default) the permutation statistic is a
regression-projection norm; when respect_SStype = TRUE it is an SS-type-aligned
QR contrast norm. Neither is identical to the fitted-model REML SSQ values reported in
object$ssq, because REML/ML decompositions account for the random-effect
covariance structure whereas permutation statistics do not. The p-values are still
valid under their respective null hypotheses; the SSQ values should not be compared
across the two sources.
Value
An hdanova object containing loadings, scores, explained variances, etc. The object has
associated plotting (asca_plots) and result (asca_results) functions.
Examples
# Load candies data
data(candies)
# Basic HDANOVA model with two factors
mod <- hdanova(assessment ~ candy + assessor, data=candies)
summary(mod)
Linear Mixed Model PCA
Description
This function mimics parts of the LiMM-PCA framework, combining ASCA+ and linear mixed models to analyse high-dimensional designed data. The default is to use REML estimation and scaling of the backprojected errors. See examples for alternatives.
Usage
limmpca(
formula,
data,
pca.in = 5,
aug_error = 0.05,
use_ED = FALSE,
REML = TRUE,
contrasts = "contr.sum",
permute = FALSE,
perm.type = c("approximate", "exact"),
SStype = "III",
...
)
Arguments
formula |
Model formula accepting a single response (block) and predictors. See Details for more information. |
data |
The data set to analyse. |
pca.in |
Compress response before ASCA (number of components), default = 5. |
aug_error |
Error term of model ("denominator", "residual", numeric alpha-value). The latter implies the first with a scaling factor. |
use_ED |
Use Effective Dimensions instead of degrees of freedom when scaling. |
REML |
Use restricted maximum likelihood estimation. Alternatives: TRUE (default), FALSE (ML), NULL (least squares). |
contrasts |
Effect coding: "sum" (default = sum-coding), "weighted", "reference", "treatment". |
permute |
Number of permutations to perform (default = 1000). |
perm.type |
Type of permutation to perform, either "approximate" or "exact" (default = "approximate"). |
SStype |
Type of sum-of-squares: "I" = sequential, "II" = last term, obeying marginality, "III" (default) = last term, not obeying marginality. |
... |
Additional arguments to |
Details
The Sum of Squares for the model is dependent on the SStype of the model. For SStype = "I" and SStype = "II" the SSQ is based on LLR (possibly inflating large contributions), while it is directly estimated from the model for SStype = "III". SStype = "III" is the default for LiMM-PCA and should be combined with sum coding. Sum of Squares for the random effects are based on the variance components.
Value
An object of class limmpca, inheriting from the general asca class.
References
Martin, M. and Govaerts, B. (2020). LiMM-PCA: Combining ASCA+ and linear mixed models to analyse high-dimensional designed data. Journal of Chemometrics, 34(6), e3232.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Prediction on new data: predict_asca_family.
Examples
# Load candies data
data(candies)
# Default LiMM-PCA model with two factors and interaction, 5 PCA components
mod <- limmpca(assessment ~ candy*r(assessor), data=candies)
summary(mod)
scoreplot(mod, factor = "candy")
# LiMM-PCA with least squares estimation and 8 PCA components
modLS <- limmpca(assessment ~ candy*r(assessor), data=candies, REML=NULL, pca.in=8)
summary(modLS)
scoreplot(modLS, factor = "candy")
# Load Caldana data
data(caldana)
# Combining effects in LiMM-PCA (assuming light is a random factor)
mod.comb <- limmpca(compounds ~ time + comb(r(light) + r(time:light)), data=caldana, pca.in=8)
summary(mod.comb)
Log-Likelihood for HDANOVA Objects
Description
Extract log-likelihood from fitted HDANOVA objects. For fixed-effects
and MoM ('r()' with REML = NULL) models, the log-likelihood is computed
from the residual sum of squares. For REML/ML mixed models ('r()' with
REML = TRUE or FALSE), the log-likelihood is extracted directly
from the stored lme4 model objects.
Usage
## S3 method for class 'hdanova'
logLik(object, ...)
Arguments
object |
A fitted |
... |
Reserved for generic compatibility. |
Details
The log-likelihood is computed as follows:
Fixed-effects and MoM models: The multivariate Gaussian log-likelihood is
\ell_j = -\frac{n}{2}\log(2\pi) - \frac{n}{2}\log(\sigma_j^2) - \frac{1}{2\sigma_j^2}\sum_i e_{ij}^2
where e_{ij} is the residual for observation i and response j,
and \sigma_j^2 = SSE_j / n is the ML variance estimate for response
column j. The total log-likelihood is \ell = \sum_j \ell_j.
REML/ML mixed models: The log-likelihood is the sum of the individual response-specific log-likelihoods from the fitted lme4 models,
\ell = \sum_j \ell_j
where \ell_j is the log-likelihood of the lme4::lmerMod fit for response j.
Value
An object of class logLik with the computed log-likelihood value,
degrees of freedom ('df'), and the number of observations ('nobs').
See Also
Model fitting: hdanova.
Information criteria: AIC and BIC.
Examples
data(candies)
mod <- hdanova(assessment ~ candy + assessor, data = candies)
ll <- logLik(mod)
print(ll)
## Not run:
# For mixed models:
mod_reml <- hdanova(assessment ~ candy + r(assessor), data = candies, REML = TRUE)
ll_reml <- logLik(mod_reml)
print(ll_reml)
## End(Not run)
Model Frame and Model Matrix for ASCA-like Models
Description
Extraction functions to retrieve the model.frame and model.matrix of an asca object.
Usage
## S3 method for class 'asca'
model.frame(formula, ...)
## S3 method for class 'asca'
model.matrix(object, ...)
Arguments
formula |
The |
... |
Not implemented |
object |
The |
Value
A data.frame or matrix object.
See Also
Core model constructors: asca, apca,
apls, limmpca, msca and
pcanova.
Examples
# Load candies data
data(candies)
# Basic ASCA model with two factors
mod <- asca(assessment ~ candy + assessor, data=candies)
# Extract model frame and model matrix
mf <- model.frame(mod)
head(mf)
mm <- model.matrix(mod)
par.old <- par(mar=c(3,3,3,1), mgp=c(1,0.7,0))
image(t(mm[seq(165,1,-1),]), main="Model Matrix", xlab="dummy values", ylab="samples",
axes=FALSE)
par(par.old)
Multilevel Simultaneous Component Analysis - MSCA
Description
This MSCA implementation assumes a single factor to be used as between-individuals factor.
Usage
msca(
formula,
data,
contrasts = "contr.sum",
permute = FALSE,
perm.type = c("approximate", "exact"),
...
)
Arguments
formula |
Model formula accepting a single response (block) and predictors. See Details for more information. |
data |
The data set to analyse. |
contrasts |
Effect coding: "sum" (default = sum-coding), "weighted", "reference", "treatment". |
permute |
Number of permutations to perform (default = 1000). |
perm.type |
Type of permutation to perform, either "approximate" or "exact" (default = "approximate"). |
... |
Additional arguments to |
Value
An asca object containing loadings, scores, explained variances, etc. The object has
associated plotting (asca_plots) and result (asca_results) functions.
References
Smilde, A., Jansen, J., Hoefsloot, H., Lamers,R., Van Der Greef, J., and Timmerman, M.(2005). ANOVA-Simultaneous Component Analysis (ASCA): A new tool for analyzing designed metabolomics data. Bioinformatics, 21(13), 3043–3048.
Liland, K.H., Smilde, A., Marini, F., and Næs,T. (2018). Confidence ellipsoids for ASCA models based on multivariate regression theory. Journal of Chemometrics, 32(e2990), 1–13.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# Load candies data
data(candies)
# Basic MSCA model with a single factor
mod <- msca(assessment ~ candy, data=candies)
print(mod)
summary(mod)
# Result plotting for first factor
loadingplot(mod, scatter=TRUE, labels="names")
scoreplot(mod)
# Within scores
scoreplot(mod, factor="within")
# Within scores per factor level
par.old <- par(mfrow=c(3,2), mar=c(4,4,2,1), mgp=c(2,0.7,0))
for(i in 1:length(mod$scores.within))
scoreplot(mod, factor="within", within_level=i,
main=paste0("Level: ", names(mod$scores.within)[i]),
panel.first=abline(v=0,h=0,col="gray",lty=2))
par(par.old)
# Permutation testing
mod.perm <- asca(assessment ~ candy * assessor, data=candies, permute=TRUE)
summary(mod.perm)
Principal Components Analysis of Variance Simultaneous Component Analysis - PC-ANOVA
Description
This is a quite general and flexible implementation of PC-ANOVA.
Usage
pcanova(formula, data, ncomp = 0.9, contrasts = "contr.sum", ...)
Arguments
formula |
Model formula accepting a single response (block) and predictor names separated by + signs. |
data |
The data set to analyse. |
ncomp |
The number of components to retain, proportion of variation or default = minimum cross-validation error. |
contrasts |
Effect coding: "sum" (default = sum-coding), "weighted", "reference", "treatment". |
... |
Additional parameters for the |
Details
PC-ANOVA works in the opposite order of ASCA. First the response matrix is decomposed using ANOVA. Then the components are analysed using ANOVA with respect to a design or grouping in the data. The latter can be ordinary fixed effects modelling or mixed models.
Value
A pcanova object containing loadings, scores, explained variances, etc. The object has
associated plotting (pcanova_plots) and result (pcanova_results) functions.
References
Luciano G, Næs T. Interpreting sensory data by combining principal component analysis and analysis of variance. Food Qual Prefer. 2009;20(3):167-175.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# Load candies data
data(candies)
# Basic PC-ANOVA model with two factors, cross-validated opt. of #components
mod <- pcanova(assessment ~ candy + assessor, data = candies)
print(mod)
# PC-ANOVA model with interaction, minimum 90% explained variance
mod <- pcanova(assessment ~ candy * assessor, data = candies, ncomp = 0.9)
print(mod)
summary(mod)
# Tukey group letters for 'candy' per component
lapply(mod$models, function(x)
mixlm::cld(mixlm::simple.glht(x,
effect = "candy")))
# Result plotting
loadingplot(mod, scatter=TRUE, labels="names")
scoreplot(mod)
# Mixed Model PC-ANOVA, random assessor
mod.mix <- pcanova(assessment ~ candy + r(assessor), data=candies, ncomp = 0.9)
scoreplot(mod.mix)
# Fixed effects
summary(mod.mix)
PC-ANOVA Result Methods
Description
Various plotting procedures for pcanova objects.
Usage
## S3 method for class 'pcanova'
scoreplot(object, factor = 1, comps = 1:2, col = "factor", ...)
Arguments
object |
|
factor |
|
comps |
|
col |
|
... |
additional arguments to underlying methods. |
Details
Usage of the functions are shown using generics in the examples in pcanova.
Plot routines are available as
scoreplot.pcanova and loadingplot.pcanova.
Value
The plotting routines have no return.
References
Luciano G, Næs T. Interpreting sensory data by combining principal component analysis and analysis of variance. Food Qual Prefer. 2009;20(3):167-175.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# For end-to-end examples using scoreplot()/loadingplot() on PC-ANOVA
# objects, see ?pcanova.
PC-ANOVA Result Methods
Description
Standard result computation and extraction functions for ASCA (pcanova).
Usage
## S3 method for class 'pcanova'
summary(object, ...)
## S3 method for class 'summary.pcanova'
print(x, digits = 2, ...)
## S3 method for class 'pcanova'
print(x, ...)
## S3 method for class 'pcanova'
summary(object, ...)
Arguments
object |
|
... |
additional arguments to underlying methods. |
x |
|
digits |
|
Details
Usage of the functions are shown using generics in the examples in pcanova.
Explained variances are available (block-wise and global) through blockexpl and print.rosaexpl.
Object printing and summary are available through:
print.pcanova and summary.pcanova.
Scores and loadings have their own extensions of scores() and loadings() through
scores.pcanova and loadings.pcanova. Special to ASCA is that scores are on a
factor level basis, while back-projected samples have their own function in projections.pcanova.
Value
Returns depend on method used, e.g. projections.pcanova returns projected samples,
scores.pcanova return scores, while print and summary methods return the object invisibly.
References
Luciano G, Næs T. Interpreting sensory data by combining principal component analysis and analysis of variance. Food Qual Prefer. 2009;20(3):167-175.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
# For examples using print()/summary() for PC-ANOVA objects, see
# ?pcanova.
Permutation Based MANOVA - PERMANOVA
Description
Wrapper for the adonis2 function to allow ordinary formula input.
Usage
permanova(formula, data, ...)
Arguments
formula |
Model formula accepting a single response matrix and predictors.
See details in |
data |
The data set to analyse. |
... |
Additional arguments to |
Value
An ANOVA table with permutation-based p-values.
See Also
Related wrappers and model families: prc,
asca, apca, apls,
limmpca, msca and pcanova.
Examples
data(caldana)
(pr <- permanova(compounds ~ light * time, caldana))
Permutation for SS-Type-Aligned HDANOVA
Description
Permutation testing for hdanova() objects using SS-type-aligned QR
effect matrices. This function is intended as an opt-in alternative to the
legacy regression-based permutation() workflow.
Usage
permutation(
object,
permute = 1000,
perm.type = c("approximate", "exact"),
respect_SStype = NULL,
unique.digits = 12,
unique.frac = 0.95,
exhaustive.warn = TRUE
)
Arguments
object |
A |
permute |
Number of permutations to perform (default = 1000). |
perm.type |
Type of permutation to perform, either |
respect_SStype |
Logical or |
unique.digits |
Number of digits used when rounding permutation SSQ values before
checking uniqueness (default = 12). Set to |
unique.frac |
Minimum fraction of unique rounded SSQ values required to avoid
warning (default = 0.95). Set to |
exhaustive.warn |
Logical; if |
Details
The permutation statistics are computed from SS-type-aligned QR reduced/full
model contrasts rather than the legacy regression-based LS matrices. Fixed
models and mixed MoM models are supported. Approximate permutation uses a
relaxed global shuffle of observations; exact permutation uses permissible
block-restricted shuffles. For REML/ML mixed models with
respect_SStype = TRUE, a warning is issued to highlight that
permutation statistics and fitted-model REML SSQ decompositions are based on
different computational definitions.
Value
An updated hdanova object with permute results.
Partial Least Squares (PLS) for HDANOVA
Description
This function performs Partial Least Squares (PLS) on a hdanova.
Usage
pls(object, ...)
## Default S3 method:
pls(object, ...)
Arguments
object |
A |
... |
Additional arguments (not used). |
Details
For residuals, PCA is performed instead of PLS as there is no natural response.
Value
An updated hdanova object with PLS results.
See Also
Main wrapper: apls.
Related decomposition: sca.
Plotting and summaries: asca_plots and asca_results.
Examples
# Load candies data
data(candies)
# Basic HDANOVA model with two factors
mod <- hdanova(assessment ~ candy + assessor, data=candies)
mod <- pls(mod)
scoreplot(mod)
Principal Response Curves
Description
Wrapper for the prc function to allow for formula input.
Usage
prc(formula, data, ...)
Arguments
formula |
Model formula accepting a single response (block) and predictors. If no predictor is called 'time', time is assumed to be the second predictor. |
data |
The data set to analyse. |
... |
Additional arguments to |
Value
An object of class prc.
See Also
Main methods: asca, apca, limmpca, msca, pcanova, prc and permanova.
Workhorse function underpinning most methods: hdanova.
Extraction of results and plotting: asca_results, asca_plots, pcanova_results and pcanova_plots
Examples
data(caldana)
(pr <- prc(compounds ~ light * time, caldana))
summary(pr)
Predict for HDANOVA Objects
Description
Reconstructs an HDANOVA-style object on newdata without refitting by
reusing stored coefficients and projection objects from the fitted model.
This implementation supports fixed-effects, mixed MoM (r() with
REML = NULL), and REML/ML mixed workflows (r() with
REML = TRUE or FALSE).
Usage
## S3 method for class 'hdanova'
predict(object, newdata, ...)
Arguments
object |
A fitted |
newdata |
A data frame containing the variables used in the original model formula. |
... |
Reserved for generic compatibility; runtime overrides are not supported. |
Value
An hdanova-family object computed on newdata.
See Also
ASCA-family prediction wrappers: predict_asca_family.
Model constructors: hdanova, asca,
apca, apls, limmpca,
msca and pcanova.
Examples
data(candies)
# Train/test split (every third sample to test)
test_idx <- seq(3, nrow(candies), by = 3)
train_idx <- setdiff(1:nrow(candies),test_idx)
candies_train <- candies[train_idx, ]
candies_test <- candies[test_idx, ]
# Fixed-effects model prediction
mod <- hdanova(assessment ~ candy + assessor, data = candies_train)
pred <- predict(mod, newdata = candies_test)
var_idx <- seq_len(ncol(mod$LS$candy))
old.par <- par(mfrow = c(1,2), mar = c(4,4,2,1), mgp = c(2,0.7,0))
image(x = var_idx, y = seq_along(train_idx), z = t(mod$LS$candy),
xaxt = "n", yaxt = "n", main = "Original candy LS",
xlab = "Variable index", ylab = "Train sample index")
axis(1, at = var_idx, labels = var_idx)
axis(2, at = seq_along(train_idx), labels = train_idx)
image(x = var_idx, y = seq_along(test_idx), z = t(pred$LS$candy),
xaxt = "n", yaxt = "n", main = "Predicted candy LS",
xlab = "Variable index", ylab = "Test sample index")
axis(1, at = var_idx, labels = var_idx)
axis(2, at = seq_along(test_idx), labels = test_idx)
par(old.par)
# Mixed MoM model prediction (r() with REML = NULL)
mod_mom <- hdanova(assessment ~ candy + r(assessor), data = candies_train)
pred_mom <- predict(mod_mom, newdata = candies_test)
cat("Mixed MoM model prediction successful.\n")
cat("SSQ names:", paste(names(pred_mom$ssq), collapse = "|"), "\n")
cat("dfDenom:", paste(pred_mom$dfDenom, collapse = "|"), "\n")
# REML mixed model prediction (r() with REML = TRUE)
mod_reml <- hdanova(assessment ~ candy + r(assessor), data = candies_train, REML = TRUE)
pred_reml <- predict(mod_reml, newdata = candies_test)
cat("REML mixed model prediction successful.\n")
cat("SSQ names:", paste(names(pred_reml$ssq), collapse = "|"), "\n")
cat("dfDenom:", paste(pred_reml$dfDenom, collapse = "|"), "\n")
Predict Methods for ASCA-family Objects
Description
Reconstructs the HDANOVA base on newdata via predict.hdanova()
and then applies the class-specific decomposition step:
sca() for ASCA/APCA/MSCA and pls() for APLS.
By default, decomposition is done by projection onto training component
spaces. A refit mode is also available.
Usage
## S3 method for class 'asca'
predict(object, newdata, decomposition = c("project", "refit"), ...)
## S3 method for class 'apca'
predict(object, newdata, decomposition = c("project", "refit"), ...)
## S3 method for class 'msca'
predict(object, newdata, decomposition = c("project", "refit"), ...)
## S3 method for class 'apls'
predict(object, newdata, decomposition = c("project", "refit"), ...)
## S3 method for class 'limmpca'
predict(object, newdata, decomposition = c("project", "refit"), ...)
Arguments
object |
A fitted |
newdata |
A data frame containing variables from the original model formula. |
decomposition |
Decomposition mode: |
... |
Reserved for compatibility; forwarded to |
Value
A predicted object of the same high-level class as object.
See Also
Base prediction engine: predict.hdanova.
Related model constructors: asca, apca,
apls, msca and limmpca.
Examples
data(candies)
test_idx <- seq(3, nrow(candies), by = 3)
train_idx <- setdiff(seq_len(nrow(candies)), test_idx)
candies_train <- candies[train_idx, ]
candies_test <- candies[test_idx, ]
mod_asca <- asca(assessment ~ candy * assessor, data = candies_train)
pred_asca <- predict(mod_asca, newdata = candies_test)
scoreplot(mod_asca, factor="candy", legend=TRUE)
with(pred_asca$projected, points(candy[,1], candy[,2], pch="x", cex=0.8,
col=as.numeric(pred_asca$model.frame$candy)))
pred_asca_refit <- predict(mod_asca, newdata = candies_test, decomposition = "refit")
mod_apca <- apca(assessment ~ candy + assessor, data = candies_train)
pred_apca <- predict(mod_apca, newdata = candies_test)
mod_msca <- msca(assessment ~ candy, data = candies_train)
pred_msca <- predict(mod_msca, newdata = candies_test)
mod_apls <- apls(assessment ~ candy + assessor, data = candies_train)
pred_apls <- predict(mod_apls, newdata = candies_test)
mod_limmpca <- limmpca(assessment ~ candy + r(assessor),
data = candies_train, pca.in = 3)
pred_limmpca <- predict(mod_limmpca, newdata = candies_test)
Objects exported from other packages
Description
These objects are imported from other packages. Follow the links below to see their documentation.
- pls
MSEP(),R2(),RMSEP(),coefplot(),cvsegments(),loading.weights(),loadingplot(),loadings(),mvrValstats(),pcr(),plsr(),predplot(),scoreplot(),scores(),validationplot()
Rotation test for HDANOVA
Description
Rotation testing for HDANOVA. This function performs
random orthogonal rotations of exchangeable residual units for each approved
effect and adds the resulting null distributions to the hdanova object.
Usage
rotation(
object,
rotate = 1000,
unique.digits = 12,
unique.frac = 0.95,
block.type = c("denominator", "global")
)
Arguments
object |
A |
rotate |
Number of random rotations to perform (default = 1000). |
unique.digits |
Number of digits used when rounding rotation SSQ values before
checking uniqueness (default = 12). Set to |
unique.frac |
Minimum fraction of unique rounded SSQ values required to avoid
warning (default = 0.95). Set to |
block.type |
Rotation blocking strategy. |
Value
An updated hdanova object with rotation-test results stored in
object$permute for compatibility with existing summary and plotting tools.
See Also
Base fitting: hdanova.
Permutation alternative: permutation.
Plot helper: rotationplot.
Examples
# Load candies data
data(candies)
# Basic HDANOVA model with two factors
mod <- hdanova(assessment ~ candy + assessor, data=candies)
# Rotation test
modRot <- rotation(mod)
summary(modRot)
# Plot null distribution for "candy" effect
rotationplot(modRot, factor="candy")
Simultaneous Component Analysis
Description
This function performs Simultaneous Component Analysis (SCA) on a hdanova object.
Usage
sca(object)
Arguments
object |
A |
Value
An updated hdanova object with SCA results.
See Also
Model constructors and wrappers: hdanova,
asca, apca, limmpca and
msca.
Plotting and summaries: asca_plots and asca_results.
Examples
# Load candies data
data(candies)
# Basic HDANOVA model with two factors
mod <- hdanova(assessment ~ candy + assessor, data=candies)
mod <- sca(mod)
scoreplot(mod)
Flip signs of a component/factor combination in a SCA/PCA object
Description
This function flips the sign of a selected component in a selected factor of an asca object.
This affects both scores, loadings and projected data.
Usage
signflip(object, factor, comp)
Arguments
object |
|
factor |
|
comp |
|
Value
An asca object with the sign of the selected component flipped.
Examples
# Load candies data
data(candies)
# Basic HDANOVA model with two factors
mod <- hdanova(assessment ~ candy + assessor, data=candies)
mod <- sca(mod)
old.par <- par(mfrow=c(1,2), mar=c(4,4,1,1))
scoreplot(mod, factor="candy")
loadingplot(mod, factor="candy")
par(old.par)
# Flip the sign of the first component of the candy factor
mod <- signflip(mod, factor="candy", comp=1)
old.par <- par(mfrow=c(1,2), mar=c(4,4,1,1))
scoreplot(mod, factor="candy")
loadingplot(mod, factor="candy")
par(old.par)
Timeplot for Combined Effects
Description
Timeplot for Combined Effects
Usage
timeplot(
object,
factor,
time,
comb,
comp = 1,
ylim,
x_time = FALSE,
xlab = time,
ylab = paste0("Score ", comp),
lwd = 2,
...
)
Arguments
object |
|
factor |
|
time |
|
comb |
|
comp |
|
ylim |
|
x_time |
|
xlab |
|
ylab |
|
lwd |
|
... |
additional arguments to |
Value
Nothing
Examples
data("caldana")
mod.comb <- asca(compounds ~ time + comb(light + time:light), data=caldana)
# Default time axis
timeplot(mod.comb, factor="light", time="time", comb=2)
# Non-equispaced time axis (using time levels)
timeplot(mod.comb, factor="light", time="time", comb=2, x_time=TRUE)
# Second component
timeplot(mod.comb, factor="light", time="time", comb=2, comp=2, x_time=TRUE)
Update a Model without Factor
Description
Perform a model update while removing a chosen factor. Hierarchical corresponds to type "II" sum-of-squares, i.e., obeying marginality, while non-hierarchical corresponds to type "III" sum-of-squares.
Usage
update_without_factor(model, fac, hierarchical = TRUE)
Arguments
model |
|
fac |
|
hierarchical |
|
Value
An updated model object is returned. If the supplied model is of
type lmerMod and no random effects are left, the model is
automatically converted to a linear model before updating.