,
Hanying Jiang [aut]| Title: | Multimodal Mediation Analysis |
|---|---|
| Description: | Multimodal mediation analysis is an emerging problem in microbiome data analysis. Multimedia make advanced mediation analysis techniques easy to use, ensuring that all statistical components are transparent and adaptable to specific problem contexts. The package provides a uniform interface to direct and indirect effect estimation, synthetic null hypothesis testing, bootstrap confidence interval construction, and sensitivity analysis. More details are available in Jiang et al. (2024) "multimedia: Multimodal Mediation Analysis of Microbiome Data" <doi:10.1101/2024.03.27.587024>. |
| Authors: | Kris Sankaran [aut, cre] ,
Hanying Jiang [aut] |
| Maintainer: | Kris Sankaran <[email protected]> |
| License: | CC0 |
| Version: | 0.2.1 |
| Built: | 2025-02-15 06:48:35 UTC |
| Source: | https://github.com/krisrs1128/multimedia |
We can subset samples by indexing into a mediation dataset. It will subsample all fields – pretreatments, treatments, mediators, and outcomes. Note that there is no way to subset columns in this way, since they would be different across each source.
## S4 method for signature 'mediation_data,ANY,ANY,ANY' x[i, j, ..., drop = TRUE]## S4 method for signature 'mediation_data,ANY,ANY,ANY' x[i, j, ..., drop = TRUE]
x |
An object of class |
i |
An integer or integer/logical vector specifying the samples to subset to. |
j |
A placeholder to argree with R's |
... |
A placeholder to agree with R's |
drop |
A placeholder to agree with R's |
A version of the input mediation_data object whose @mediators,
@outcomes, @treatments, and @pretreatments rows have all been subsetted
according to i.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) exper[1] exper[1:10]exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) exper[1] exper[1:10]
Helper function for printing ANSI in Rmarkdown output. Use this at the start of your Rmarkdown files to include colors in the printed object names in the final compiled output.
ansi_aware_handler(x, options)ansi_aware_handler(x, options)
x |
A character vector potentially including ANSI. |
options |
Unused placeholder argument. |
Taken from the post at
https://blog.djnavarro.net/posts/2021-04-18_pretty-little-clis/
A string with HTML reformatted to ensure colors appear in printed code blocks in rmarkdown output.
knitr::knit_hooks$set(output = ansi_aware_handler) options(crayon.enabled = TRUE)knitr::knit_hooks$set(output = ansi_aware_handler) options(crayon.enabled = TRUE)
mediation_data to a single data.frameIt can be helpful to combine all the data in a mediation_data S4 object
into a single data.frame. This concatenates all data sources horizontally,
into an N samples x (all outcomes, mediators, treatments, ...) matrix.
bind_mediation(exper)bind_mediation(exper)
exper |
An object of class |
A data.frame containing all variables from the original
mediation_data object.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) exper bind_mediation(exper) exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") exper bind_mediation(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) exper bind_mediation(exper) exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") exper bind_mediation(exper)
Given a mediation model specification, estimators fs, and original dataset exper, this will re-estimate the mediation model on resampled versions of exper and apply each estimator in fs to construct bootstrap distributions associated wtih those estimators.
bootstrap(model, exper, fs = NULL, B = 1000, progress = TRUE)bootstrap(model, exper, fs = NULL, B = 1000, progress = TRUE)
model |
An object of class multimedia with specified mediation and outcome models that we want to re-estimate across B bootstrap samples. |
exper |
An object of class multimedia_data containing the mediation and outcome data from which the direct effects are to be estimated. |
fs |
The estimators whose bootstrap samples we are interested in. These are assumed to be a vector of functions (for example, direct_effect or indirect_effect), and they will each be applied to each bootstrap resample. |
B |
The number of bootstrap samples. Defaults to 1000. |
progress |
A logical indicating whether to show a progress bar. |
stats A list of length B containing the results of the fs applied on each of the B bootstrap resamples.
# example with null data. We set B to 5 just to execute quickly -- it's not # actually a practical choice of B exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> bootstrap(exper, B = 5) # example with another dataset exper <- demo_spline(n_samples = 100, tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") samples <- multimedia(exper, rf_model(num.trees = 1e3)) |> bootstrap(exper, B = 5) ggplot2::ggplot(samples$direct_effect) + ggplot2::geom_histogram( ggplot2::aes(direct_effect, fill = indirect_setting), bins = 15 ) + ggplot2::facet_wrap(~outcome, scales = "free")# example with null data. We set B to 5 just to execute quickly -- it's not # actually a practical choice of B exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> bootstrap(exper, B = 5) # example with another dataset exper <- demo_spline(n_samples = 100, tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") samples <- multimedia(exper, rf_model(num.trees = 1e3)) |> bootstrap(exper, B = 5) ggplot2::ggplot(samples$direct_effect) + ggplot2::geom_histogram( ggplot2::aes(direct_effect, fill = indirect_setting), bins = 15 ) + ggplot2::facet_wrap(~outcome, scales = "free")
Apply a Bayesian regression model in parallel across each response $y$ in an outcome or mediation model. This can be helpful when we want to share information across related
brms_model(...)brms_model(...)
... |
Keyword parameters passed to brm. |
model An object of class model with estimator, predictor, and
sampler functions associated wtih a Bayesian regression model.
glmnet_model lnm_model rf_model lm_model
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper, brms_model()) # call estimate() on this to fitexper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper, brms_model()) # call estimate() on this to fit
This samples from the posterior predictive for each component in a multiresponse Bayesian Regression model.
brms_sampler(fits, newdata = NULL, indices = NULL, ...)brms_sampler(fits, newdata = NULL, indices = NULL, ...)
fits |
The fitted 'BRMS' model model from which to draw samples. |
newdata |
A data.frame containing new inputs from which to sample responses. If NULL, defaults to the data used to estimate fit. |
indices |
The coordinates of the response from which we want to sample. |
... |
Additional arguments to pass to |
A data.frame containing a single posterior predictive sample at each of the newdata rows passed into a fitted BRMS model. Each column corresponds to one outcome variable, each row to the associated row in the newdata input..
Given a fitted multimedia model, contrast the mediation and outcome predictions associated wtih two treatment profiles.
contrast_predictions(model, profile1, profile2, ...)contrast_predictions(model, profile1, profile2, ...)
model |
An object of class multimedia containing the estimated mediation and outcome models whose mediation and outcome predictions we want to compare. |
profile1 |
An object of class |
profile2 |
An object of class |
... |
Additional arguments to pass to |
A list with two elements, mediators and outcomes, containing the
differences in the predicted M(T') - M(T) and Y(T', M(T')) - Y(T, M(T))
between the two profiles T and T'.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) model <- multimedia(exper) |> estimate(exper) t1 <- data.frame(treatment = factor("Treatment")) t2 <- data.frame(treatment = factor("Control")) profile1 <- setup_profile(model, t1, t1) profile2 <- setup_profile(model, t2, t2) contrast_predictions(model, profile1, profile2)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) model <- multimedia(exper) |> estimate(exper) t1 <- data.frame(treatment = factor("Treatment")) t2 <- data.frame(treatment = factor("Control")) profile1 <- setup_profile(model, t1, t1) profile2 <- setup_profile(model, t2, t2) contrast_predictions(model, profile1, profile2)
Given a fitted multimedia model, contrast sampled mediation and outcome data associated with two treatment profiles.
contrast_samples(model, profile1, profile2, ...)contrast_samples(model, profile1, profile2, ...)
model |
An object of class multimedia containing the estimated mediation and outcome models whose mediation and outcome predictions we want to compare. |
profile1 |
An object of class |
profile2 |
An object of class |
... |
Additional arguments to pass to |
A list with two elements, mediators and outcomes, containing the
differences in the sampled M(T') - M(T) and Y(T', M(T')) - Y(T, M(T))
between the two profiles T and T'.
A list with two elements, mediators and outcomes. These contrast the values of the mediator and outcomes under the two profiles T and T'.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) model <- multimedia(exper) |> estimate(exper) t1 <- data.frame(treatment = factor("Treatment")) t2 <- data.frame(treatment = factor("Control")) profile1 <- setup_profile(model, t1, t1) profile2 <- setup_profile(model, t2, t2) contrast_samples(model, profile1, profile2) samples <- purrr::map( seq_len(100), ~ contrast_samples(model, profile1, profile2) ) hist(sapply(samples, \(x) x[[1]]$ASV1)) hist(sapply(samples, \(x) x[[1]]$ASV2))exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) model <- multimedia(exper) |> estimate(exper) t1 <- data.frame(treatment = factor("Treatment")) t2 <- data.frame(treatment = factor("Control")) profile1 <- setup_profile(model, t1, t1) profile2 <- setup_profile(model, t2, t2) contrast_samples(model, profile1, profile2) samples <- purrr::map( seq_len(100), ~ contrast_samples(model, profile1, profile2) ) hist(sapply(samples, \(x) x[[1]]$ASV1)) hist(sapply(samples, \(x) x[[1]]$ASV2))
This is a simple dataset with random data, used simply to illustrate the design of multimedia. There is no real association between any treatments, mediators, or outcomes. It always returns a single outcome (PHQ) SummarizedExperiment, and it randomly assigns samples to Treatment and Control (see the colData). It stores all the hypothetical mediator data in the assay slot.
demo_joy(n_samples = 100, n_mediators = 5, n_pretreatment = 3)demo_joy(n_samples = 100, n_mediators = 5, n_pretreatment = 3)
n_samples |
How many random samples to generate? |
n_mediators |
How many random mediators to generate? |
n_pretreatment |
How many random pretreatment variables? |
SE The summarized experiment containing random data.
demo_joy() demo_joy(n_samples = 2, n_mediators = 20)demo_joy() demo_joy(n_samples = 2, n_mediators = 20)
This is a simple dataset with nonlinear relationships between the outcome and mediators. It is used simply to illustrate the design of multimedia. The mediator->outcome effect is generated from a random spline function.
demo_spline(n_samples = 5000, tau = c(2, 2))demo_spline(n_samples = 5000, tau = c(2, 2))
n_samples |
The number of samples to generate in the toy example |
tau |
The true direct effects associated with the two outcomes. Defaults to 2, 2. |
xy A data.frame whose columns include the treatment, mediation, and outcome variables.
demo_spline()demo_spline()
Estimate direct effects associated with a multimedia model. These estimates
are formed using Equation (10) of our paper. Rather than providing this
average, this function returns the estimated difference for each $j$. To
average across all j, this result can be passed to the ' effect_summary
function.
direct_effect(model, exper = NULL, t1 = 1, t2 = 2)direct_effect(model, exper = NULL, t1 = 1, t2 = 2)
model |
An object of class multimedia containing the estimated mediation and outcome models whose mediation and outcome predictions we want to compare. |
exper |
An object of class multimedia_data containing the mediation and outcome data from which the direct effects are to be estimated. |
t1 |
The reference level of the treatment to be used when computing the direct effect. |
t2 |
The alternative level of the treatment to be used when computing the direct effect. |
A data.frame summarizing the direct effects associated with different settings of j in the equation above.
effect_summary
# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) fit <- multimedia(exper) |> estimate(exper) direct_effect(fit) direct_effect(fit, t1 = 2, t2 = 1) direct_effect(fit, t1 = 2, t2 = 2) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) direct_effect(fit)# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) fit <- multimedia(exper) |> estimate(exper) direct_effect(fit) direct_effect(fit, t1 = 2, t2 = 1) direct_effect(fit, t1 = 2, t2 = 2) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) direct_effect(fit)
We often want to access the DAG for different mediation-related S4 objects. This generic helps us access these graphical model edges lists. See method instantiations for specific examples.
edges(object)edges(object)
object |
An object whose DAG structure we want to access. |
The output depends on the S4 object that is passed. For multimedia objects, this will return an edgelist as a two column data.frame.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> edges()exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> edges()
This is an accessor to the edges slot in a multimedia object. It is the internal representation of the variable conditional dependence graph encoded by the mediation model's DAG.
## S4 method for signature 'multimedia' edges(object)## S4 method for signature 'multimedia' edges(object)
object |
An object of class multimedia. |
A data.frame whose rows give edges in the mediation analysis DAG.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> edges()exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> edges()
This averages direct or indirect effects across settings j, leading to the effect estimates given in equation (10) of the preprint.
effect_summary(effects)effect_summary(effects)
effects |
The output from direct_effect or indirect_effect. A data.frame containing effect estimates for each variable and indirect/direct setting along rows. |
A version of the input with all indirect/direct settings averaged over.
direct_effect indirect_effect
# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> estimate(exper) |> direct_effect() |> effect_summary() # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> estimate(exper) |> direct_effect() |> effect_summary()# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> estimate(exper) |> direct_effect() |> effect_summary() # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> estimate(exper) |> direct_effect() |> effect_summary()
estimate provides a unified interface to estimate all the models that can
be encapsulated within a multimedia class. It simply calls the
multimedia object. The resulting estimates can be used for downstream direct
effect estimation.
estimate(model, exper)estimate(model, exper)
model |
An object of class multimedia containing the estimated mediation and outcome models whose mediation and outcome predictions we want to compare. |
exper |
An object of class multimedia_data containing the mediation and outcome data from which the direct effects are to be estimated. |
A version of the input modified in place so that the @estimates slot has been filled.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> estimate(exper) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> estimate(exper) # example with another model exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper, glmnet_model()) |> estimate(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> estimate(exper) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> estimate(exper) # example with another model exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper, glmnet_model()) |> estimate(exper)
This defines a generic estimator function, which can be applied to different multimedia model objects. It creates a unified interface to estimating diverse mediation and outcome model families.
estimator(object) estimator(object)estimator(object) estimator(object)
object |
A model object that we want to estimate. |
A function that can be called with formula and data arguments, like
A fitted version of the input model class.
m <- lm_model() estimator(m)(mpg ~ hp + wt, data = mtcars) m <- rf_model() fit <- estimator(m)(mpg ~ hp + wt, data = mtcars)m <- lm_model() estimator(m)(mpg ~ hp + wt, data = mtcars) m <- rf_model() fit <- estimator(m)(mpg ~ hp + wt, data = mtcars)
Accessor for Estimators
## S4 method for signature 'model' estimator(object)## S4 method for signature 'model' estimator(object)
object |
An object of class |
A fitted version of the input model class.
m <- rf_model() fit <- estimator(m)(mpg ~ hp + wt, data = mtcars)m <- rf_model() fit <- estimator(m)(mpg ~ hp + wt, data = mtcars)
This is a small helepr function to concatenate the colData and assays within SummarizedExperiment objects into a single data.frame. This is not necessary for the essential multimedia workflow, it is only exported for potential independent interest.
exper_df(exper)exper_df(exper)
exper |
An object of class SummarizedExperiment. |
A data.frame combining all slots of a multimedia object.
demo_joy() |> exper_df()demo_joy() |> exper_df()
This function computes a threshold for indirect or direct effect estimates that controls the false discovery rate according to estimates made using real and synthetic null data, against the null hypotheses that effects are zero. It computes the proportion of synthetic null estimates that are among the top K largest effects (in magnitude) as an estimate of the FDR.
fdr_summary(contrast, effect = "indirect_overall", q_value = 0.15)fdr_summary(contrast, effect = "indirect_overall", q_value = 0.15)
contrast |
A data.frame summarizing the differences between outcomes
across hypothetical treatments, typically as output by |
effect |
Either "indirect_overall" (the default), "indirect_pathwise", or "direct_effect" specifying the type of effect that we want to select. |
q_value |
The target for false discovery rate control. The last time the estimated FDR is above this threshold is smallest magnitude of effect size that we will consider. |
fdr A data.frame specifying, for each candidate effect, whether it should be selected.
# example with null data - notice synthetic data has larger effect. exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> estimate(exper) |> null_contrast(exper) |> fdr_summary("direct_effect") multimedia(exper) |> estimate(exper) |> null_contrast(exper, "M->Y", indirect_overall) |> fdr_summary("indirect_overall") # example with another dataset - synthetic effect is smaller. exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> estimate(exper) |> null_contrast(exper) |> fdr_summary("direct_effect") multimedia(exper) |> estimate(exper) |> null_contrast(exper, "M->Y", indirect_overall) |> fdr_summary("indirect_overall")# example with null data - notice synthetic data has larger effect. exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> estimate(exper) |> null_contrast(exper) |> fdr_summary("direct_effect") multimedia(exper) |> estimate(exper) |> null_contrast(exper, "M->Y", indirect_overall) |> fdr_summary("indirect_overall") # example with another dataset - synthetic effect is smaller. exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> estimate(exper) |> null_contrast(exper) |> fdr_summary("direct_effect") multimedia(exper) |> estimate(exper) |> null_contrast(exper, "M->Y", indirect_overall) |> fdr_summary("indirect_overall")
Apply a regularized (generalized) linear model in parallel across each response $y$ in an outcome or mediation model. This can be helpful when we have many mediators or pretreatment variables, making the input high-dimensional.
glmnet_model(progress = TRUE, ...)glmnet_model(progress = TRUE, ...)
progress |
A logical indicating whether to show a progress bar during estimation. |
... |
Keyword parameters passed to package 'glmnet'. |
model An object of class model with estimator, predictor, and
sampler functions associated wtih a lienar model.
model lm_model rf_model
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper, glmnet_model(lambda = 1)) |> estimate(exper) multimedia(exper, glmnet_model(lambda = 1), glmnet_model()) |> estimate(exper) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper, glmnet_model(lambda = 0.1)) |> estimate(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper, glmnet_model(lambda = 1)) |> estimate(exper) multimedia(exper, glmnet_model(lambda = 1), glmnet_model()) |> estimate(exper) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper, glmnet_model(lambda = 0.1)) |> estimate(exper)
This assumes a continuous response, so that the out-of-sample MSE can be used to estimate the outcome variability sigma.
glmnet_sampler(fits, newdata = NULL, indices = NULL, lambda_ix = 1, ...)glmnet_sampler(fits, newdata = NULL, indices = NULL, lambda_ix = 1, ...)
fits |
The fitted 'glmnet' package model model from which to draw samples. |
newdata |
A data.frame containing new inputs from which to sample responses. If NULL, defaults to the data used to estimate fit. |
indices |
The coordinates of the response from which to draw samples. |
lambda_ix |
A regularization strength parameter used to maintain consistency with estimation. Not used during sampling. |
... |
Additional parameters to pass to predict.glmnet |
y_star A data.frame of samples y associated with the new inputs.
m <- glmnet_model() fit <- estimator(m)(mpg ~ hp + wt, data = mtcars) glmnet_sampler(fit, mtcars) plm <- parallelize(glmnetUtils::glmnet) fit <- plm(mpg + disp ~ hp + wt, data = mtcars) glmnet_sampler(fit, mtcars)m <- glmnet_model() fit <- estimator(m)(mpg ~ hp + wt, data = mtcars) glmnet_sampler(fit, mtcars) plm <- parallelize(glmnetUtils::glmnet) fit <- plm(mpg + disp ~ hp + wt, data = mtcars) glmnet_sampler(fit, mtcars)
Direct Effects from Estimated Model
indirect_overall(model, exper = NULL, t1 = 1, t2 = 2)indirect_overall(model, exper = NULL, t1 = 1, t2 = 2)
model |
An object of class multimedia containing the estimated mediation and outcome models whose mediation and outcome predictions we want to compare. |
exper |
An object of class multimedia_data containing the mediation and outcome data from which the direct effects are to be estimated. |
t1 |
The reference level of the treatment to be used when computing the indirect effect. |
t2 |
The alternative level of the treatment to be used when computing the indirect effect. |
Estimate direct effects associated with a multimedia model. These estimates are formed using Equation (10) of our preprint.
A data.frame summarizing the overall indirect effects associated with different settings of j in the equation above.
# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) fit <- multimedia(exper) |> estimate(exper) indirect_overall(fit) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) indirect_overall(fit)# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) fit <- multimedia(exper) |> estimate(exper) indirect_overall(fit) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) indirect_overall(fit)
Indirect Effects via Single Mediation Paths
indirect_pathwise(model, exper = NULL, t1 = 1, t2 = 2, progress = TRUE)indirect_pathwise(model, exper = NULL, t1 = 1, t2 = 2, progress = TRUE)
model |
An object of class multimedia containing the estimated mediation and outcome models whose mediation and outcome predictions we want to compare. |
exper |
An object of class multimedia_data containing the mediation and outcome data from which the direct effects are to be estimated. |
t1 |
The reference level of the treatment to be used when computing the (pathwise) indirect effect. |
t2 |
The alternative level of the treatment to be used when computing the (pathwise) indirect effect. |
progress |
A logical indicating whether to show a progress bar. |
A data.frame summarizing the pathwise (per-mediator) indirect effects associated with different settings of the direct effect.
# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) fit <- multimedia(exper) |> estimate(exper) indirect_pathwise(fit) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) indirect_pathwise(fit)# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) fit <- multimedia(exper) |> estimate(exper) indirect_pathwise(fit) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) indirect_pathwise(fit)
Apply a linear model in parallel across each response $y$ in an outcome or mediation model. This is often useful for mediator models with few pretreatment variables, since each input is low-dimensional, even when there are many responses.
lm_model(progress = TRUE)lm_model(progress = TRUE)
progress |
A logical indicating whether to show a progress bar during estimation. |
model An object of class model with estimator, predictor, and
sampler functions associated wtih a linear model.
model
m <- lm_model() estimator(m)(mpg ~ hp + wt, data = mtcars)m <- lm_model() estimator(m)(mpg ~ hp + wt, data = mtcars)
Draw samples from a fitted linear model.
lm_sampler(fits, newdata = NULL, indices = NULL, ...)lm_sampler(fits, newdata = NULL, indices = NULL, ...)
fits |
The fitted linear model model from which to draw samples. |
newdata |
A data.frame containing new inputs from which to sample responses. If NULL, defaults to the data used to estimate fit. |
indices |
The coordinates of the response from which to draw samples. |
... |
Additional parameters passed to lm.predict |
y_star A data.frame of samples y associated with the new inputs.
fit <- lm(mpg ~ hp + wt, data = mtcars) lm_sampler(list(f = fit)) plm <- parallelize(lm) fit <- plm(mpg + disp ~ hp + wt, data = mtcars) lm_sampler(fit)fit <- lm(mpg ~ hp + wt, data = mtcars) lm_sampler(list(f = fit)) plm <- parallelize(lm) fit <- plm(mpg + disp ~ hp + wt, data = mtcars) lm_sampler(fit)
Apply a logistic normal multinomial model to jointly model a vector of count responses $y$ in an outcome or mediation model. This is a common choice for data where the parameter of interest is the composition across responses (e.g., microbiome).
lnm_model(...)lnm_model(...)
... |
Keyword parameters passed to lnm in the 'miniLNM' package. |
model An object of class model with estimator, predictor, and
sampler functions associated wtih a linear model.
model lm_model rf_model glmnet_model brms_model
m <- lnm_model() mat <- data.frame(matrix(rpois(250, 10), 25, 10)) colnames(mat) <- paste0("y", seq_len(6)) fit <- estimator(m)(y1 + y2 + y3 + y4 ~ y5 + y6, mat)m <- lnm_model() mat <- data.frame(matrix(rpois(250, 10), 25, 10)) colnames(mat) <- paste0("y", seq_len(6)) fit <- estimator(m)(y1 + y2 + y3 + y4 ~ y5 + y6, mat)
This samples from the posterior predictive of a fitted logistic-normal multinomial model.
lnm_sampler(fit, newdata = NULL, indices = NULL, ...)lnm_sampler(fit, newdata = NULL, indices = NULL, ...)
fit |
The fitted LNM model from which to draw posterior predictive samples. |
newdata |
A data.frame containing new inputs from which to sample responses. If NULL, defaults to the data used to estimate fit. |
indices |
The coordinates of the response from which to draw samples. |
... |
Additional parameters passed to sample. |
y_star A data.frame of samples y associated wtih the new inputs.
m <- lnm_model() mat <- data.frame(matrix(rpois(250, 10), 25, 10)) colnames(mat) <- paste0("y", 1:6) fit <- estimator(m)(y1 + y2 + y3 + y4 ~ y5 + y6, mat) lnm_sampler(fit, depth = 10) lnm_sampler(fit, depth = 100)m <- lnm_model() mat <- data.frame(matrix(rpois(250, 10), 25, 10)) colnames(mat) <- paste0("y", 1:6) fit <- estimator(m)(y1 + y2 + y3 + y4 ~ y5 + y6, mat) lnm_sampler(fit, depth = 10) lnm_sampler(fit, depth = 100)
mediation_data ConstructorConvert a SummarizedExperiment, phyloseq object, or data.frame into a
mediation_data object. This conversion helps to organize the variables that
lie on different parts of the mediation analysis graph, so that they are not
all kept in a homogeneous experiment or data.frame. It's possible to specify
outcome, mediator, or treatment variables using either string vectors or
tidyselect syntax, e.g., starts_with("mediator_") will match all columns of
the input data starting with the string mediator.
mediation_data(x, outcomes, treatments, mediators, pretreatments = NULL)mediation_data(x, outcomes, treatments, mediators, pretreatments = NULL)
x |
A SummarizedExperiment, phyloseq object, or data.frame whose columns contain all the variables needed for the mediation analysis. |
outcomes |
A vector or tidyselect specification of the names of all outcome variables. |
treatments |
A vector or tidyselect specification of the names of all treatment variables. |
mediators |
A vector or tidyselect specification of the names of all mediators. |
pretreatments |
A vector containing names of all pretreatment variables. Defaults to NULL, in which case the pretreatments slot will remain empty. |
result An object of class mediation_data, with separate slots for
each of the node types in a mediation analysis diagram.
mediation_data-class
# multiple outcomes, one mediator mediation_data( demo_spline(), starts_with("outcome"), "treatment", "mediator" ) # one outcome, multiple mediators mediation_data(demo_joy(), "PHQ", "treatment", starts_with("ASV"))# multiple outcomes, one mediator mediation_data( demo_spline(), starts_with("outcome"), "treatment", "mediator" ) # one outcome, multiple mediators mediation_data(demo_joy(), "PHQ", "treatment", starts_with("ASV"))
Accessor for Outcome Models
mediation_models(object)mediation_models(object)
object |
An object of class multimedia whose outcome model estimates we would like to extract. |
A list containing all the fitted mediation models.
data(mindfulness) exper <- mediation_data( mindfulness, phyloseq::taxa_names(mindfulness), "treatment", starts_with("mediator"), "subject" ) m <- multimedia(exper) mediation_models(m)data(mindfulness) exper <- mediation_data( mindfulness, phyloseq::taxa_names(mindfulness), "treatment", starts_with("mediator"), "subject" ) m <- multimedia(exper) mediation_models(m)
This is a shorthand for accessing mediator-related slots in classes exported by the multimedia package.
mediators(object)mediators(object)
object |
An object whose mediators we want to access (either their names or values). |
The output will depend on the class of the object that is passed in. For multimedia objects, this returns a character vector of mediators. For mediation data objects, this returns the mediator mediator data.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) mediators(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) mediators(exper)
This is an accessor returns the @mediators slot in a mediation_data object.
## S4 method for signature 'mediation_data' mediators(object)## S4 method for signature 'mediation_data' mediators(object)
object |
An object of class mediation_data. |
m A data.frame whose rows are samples and columns are values of mediators across those samples.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) mediators(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) mediators(exper)
This is a helper that returns the names of the mediator variables in an object of class multimedia. It parses the graph in the DAG specifying the mediation analysis, and it returns all variables between treatment and outcome.
## S4 method for signature 'multimedia' mediators(object)## S4 method for signature 'multimedia' mediators(object)
object |
An object of class multimedia. |
m A vector of strings containing the names of all the mediators.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> mediators()exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> mediators()
This is an setter method for mediators in an S4 object, usually of class mediation_data.
mediators(object) <- valuemediators(object) <- value
object |
An object whose mediators slot to modify. |
value |
The new mediator values to set within object. |
Modifies the mediators slot of the input object in place.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) mediators(exper) <- data.frame(new_ASV = rnorm(100)) mediators(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) mediators(exper) <- data.frame(new_ASV = rnorm(100)) mediators(exper)
This is an setter method for the mediators slot in a mediation data object. It lets you supply a new mediators data.frame for the object.
## S4 replacement method for signature 'mediation_data' mediators(object) <- value## S4 replacement method for signature 'mediation_data' mediators(object) <- value
object |
An object of class mediation_data. |
value |
The new mediators values for the object. |
A version of object whose mediators slot has been replaced.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) mediators(exper) <- data.frame(m = 1:10) experexper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) mediators(exper) <- data.frame(m = 1:10) exper
Data from a study of the relationship between mindfulness and the microbiome, stored as a phyloseq object. Measurements are from 50 subjects before and after a real and active control mindfulness intervention. We are interested in changes in subject-level metadata, stored in the sample_data slot
An object of class phyloseq.
A phyloseq object with 307 samples and 55 taxa. Samples are described by 7 variables (potential mediators) in the sample_data slot. There is no associated phylogenetic tree.
data(mindfulness)data(mindfulness)
To work with many model types simultaneously, multimedia uses a model class with the necessary mediation model functionality that wraps any specific implementation. The slots below define the generally required functionality for any specific implementation.
estimatorA function that takes a formula, input data frame X, and an
response data.frame $Y$ and returns a model. For example, for the random
forest model, this is created by wrapping parallelize() on the ranger()
function for random forest estimation function using the 'ranger' package.
estimatesA list containing the estimated model.
samplerA function that supports sampling new responses from the estimated model.
model_typeA string specifying the type of model associated with the class. For example, "rf_model()" denotes a random forest model.
predictorA function that returns fitted predictions given new inputs. For example, this can be the original predict() method for a multivariate response model, or it can be a loop over predicts for each feature in the mediation or outcome model.
m <- lm_model() estimator(m)(mpg ~ hp + wt, data = mtcars) m <- rf_model() estimator(m)(mpg ~ hp + wt, data = mtcars)m <- lm_model() estimator(m)(mpg ~ hp + wt, data = mtcars) m <- rf_model() estimator(m)(mpg ~ hp + wt, data = mtcars)
multimedia Constructormultimedia objects encapsulate the model and data that underlie a mediation
analysis, together with metadata (like graph structure) that contextualize
the estimation. This function can be used to construct a new multimedia
instances from a mediation_data dataset and pair of estimators.
multimedia( mediation_data, outcome_estimator = lm_model(), mediation_estimator = lm_model() )multimedia( mediation_data, outcome_estimator = lm_model(), mediation_estimator = lm_model() )
mediation_data |
An object of class |
outcome_estimator |
An object of class |
mediation_estimator |
An object of class |
An object of class multimedia encapsulating the full mediation
model and data.
multimedia-class
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) # real data example with a pretreatment variable data(mindfulness) exper <- mediation_data( mindfulness, phyloseq::taxa_names(mindfulness), "treatment", starts_with("mediator"), "subject" ) multimedia(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) # real data example with a pretreatment variable data(mindfulness) exper <- mediation_data( mindfulness, phyloseq::taxa_names(mindfulness), "treatment", starts_with("mediator"), "subject" ) multimedia(exper)
Number of Mediators in a Multimedia Object
n_mediators(object)n_mediators(object)
object |
An object of class multimedia. |
An integer specifying the number of mediators.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> n_mediators()exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> n_mediators()
Number of Outcomes in a Multimedia Object
n_outcomes(object)n_outcomes(object)
object |
An object of class multimedia. |
An integer specifying the number of outcomes.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> n_outcomes()exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> n_outcomes()
How many samples in the mediation dataset?
## S4 method for signature 'mediation_data' nrow(x)## S4 method for signature 'mediation_data' nrow(x)
x |
The |
An integer giving the number of samples in the mediation object.
One way to calibrate our conclusions from complex workflows is to see how they would look on data where we know that there is no effect. This function compares estimators f between real and synthetic null data, where the null removes a set of edges according to the nullfication argument.
null_contrast(model, exper, nullification = "T->Y", f = direct_effect)null_contrast(model, exper, nullification = "T->Y", f = direct_effect)
model |
An object of class multimedia with specified mediation and outcome models that we want to re-estimate across B bootstrap samples. |
exper |
An object of class multimedia_data containing the mediation and outcome data from which the direct effects are to be estimated. |
nullification |
A string specifying the types of edges whose effects we want to remove in the null samples. Valid options are "T->Y" (the default), "T->M", "M->Y", which remove direct effects, treatment to mediator effects, and mediator to treatment effects, respectively. |
f |
The estimator that we want to compare between real and null data.
This is assumed to be a function taking counterfactual samples, for example
|
A data.frame containing estimates on the real and synthetic data for
every coordinate in the estimator f. The column source specifies whether
the estimate was calculated using real or synthetic null data.
null_contrast fdr_summary
# example with null data - notice synthetic data has larger effect. exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> estimate(exper) |> null_contrast(exper) # example with another dataset - synthetic effect is smaller. exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> estimate(exper) |> null_contrast(exper)# example with null data - notice synthetic data has larger effect. exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> estimate(exper) |> null_contrast(exper) # example with another dataset - synthetic effect is smaller. exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> estimate(exper) |> null_contrast(exper)
For inference, we often want to work with synthetic negative controls. One
way to define them is to specify submodels of the full mediation analysis
model. This function defines submodels by removing estimated edges according
to a prespecified vector of IDs. For example, setting nulls = "T -> Y" will
remove any direct effect when sampling or obtaining predictions for the full
mediation analysis model .
nullify(multimedia, nulls = NULL)nullify(multimedia, nulls = NULL)
multimedia |
A fitted object of class multimedia with estimates along all paths in the mediation analysis DAG. |
nulls |
A string specifying the indices of edges to ignore. "T->Y", "T->M", and "M->Y" will match all edges between treatment to outcome, treatment to mediator, etc. Otherwise, the vector of indices specifying which edges to ignore. |
multimedia A version of the input multimedia model with all edges
matching nulls removed. Enables sampling of synthetic null controls.
# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) fit <- multimedia(exper) |> estimate(exper) nullify(fit, "T->M") |> estimate(exper) |> indirect_overall() nullify(fit, "T->Y") |> estimate(exper) |> direct_effect() # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) nullify(fit, "T->M") |> estimate(exper) |> indirect_overall() nullify(fit, "T->Y") |> estimate(exper) |> direct_effect()# example with null data exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) fit <- multimedia(exper) |> estimate(exper) nullify(fit, "T->M") |> estimate(exper) |> indirect_overall() nullify(fit, "T->Y") |> estimate(exper) |> direct_effect() # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) nullify(fit, "T->M") |> estimate(exper) |> indirect_overall() nullify(fit, "T->Y") |> estimate(exper) |> direct_effect()
This is an accessor to the outcome slot of a multimedia object.
outcome_model(object)outcome_model(object)
object |
An object of class multimedia. |
NULL, if not fitted, or the model learned from the training mediation data object. For models fit in parallel across outcomes (e.g., glmnet_model()) a list of separate model objects. For models fitted jointly across outcomes (e.g., lnm_model()), a single model object of that class.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> outcome_model()exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> outcome_model()
Accessor for Outcome Models
outcome_models(object)outcome_models(object)
object |
An object of class multimedia whose outcome model estimates we would like to extract. |
A list containing all the fitted outcome models.
data(mindfulness) exper <- mediation_data( mindfulness, phyloseq::taxa_names(mindfulness), "treatment", starts_with("mediator"), "subject" ) m <- multimedia(exper) outcome_models(m)data(mindfulness) exper <- mediation_data( mindfulness, phyloseq::taxa_names(mindfulness), "treatment", starts_with("mediator"), "subject" ) m <- multimedia(exper) outcome_models(m)
This is an getter method for outcomes in an S4 object, usually of class mediation_data.
outcomes(object)outcomes(object)
object |
An object whose outcomes slot to modify. |
The output depends on the S4 class of the input object. If it is a multimedia model object, it will return a character vector of the outcome variable names. Otherwise it returns the outcome data.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) outcomes(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) outcomes(exper)
This is an accessor function to the @outcomes slot in a mediation data object. It returns the entire outcomes dataset, in contrast to outcomes() applied to a multimedia object, which only returns the names of the outcome variables.
## S4 method for signature 'mediation_data' outcomes(object)## S4 method for signature 'mediation_data' outcomes(object)
object |
An object of class mediation_data. |
A data.frame whose rows are samples and columns different outcomes.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) outcomes(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) outcomes(exper)
This is a helper that returns the names of the outcome variables in an object of class multimedia. It parses the graph in the DAG specifying the mediation analysis, and it returns all variables of node type outcome.
## S4 method for signature 'multimedia' outcomes(object)## S4 method for signature 'multimedia' outcomes(object)
object |
An object of class multimedia. |
m A vector of strings containing the names of all the outcomes.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> outcomes() exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> outcomes()exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> outcomes() exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper) |> outcomes()
Set Outcomes This is an setter method for outcomes in an S4 object, usually of class mediation_data.
outcomes(object) <- valueoutcomes(object) <- value
object |
An object whose outcomes slot to modify. |
value |
The new outcome values to set within object. |
Modifies the outcomes slot of the input object in place.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) outcomes(exper) <- data.frame(new_PHQ = rnorm(100))exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) outcomes(exper) <- data.frame(new_PHQ = rnorm(100))
This is an setter method for the outcomes slot in a mediation data object. It lets you supply a new outcomes data.frame for the object.
## S4 replacement method for signature 'mediation_data' outcomes(object) <- value## S4 replacement method for signature 'mediation_data' outcomes(object) <- value
object |
An object of class mediation_data. |
value |
The new outcome values for the object. |
A version of object whose outcomes slot has been replaced.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) outcomes(exper) <- data.frame(y = 1:10) experexper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) outcomes(exper) <- data.frame(y = 1:10) exper
For many mediation and outcome models, we simply want to apply a univariate model across all response variable. Parallelize enables this conversion. For example, applying parallelize to ranger() returns a function that estimates separate random forest models for each response on the left hand side of a formula.
parallelize(f, progress = TRUE)parallelize(f, progress = TRUE)
f |
A function for estimating a single response model given a formula and input dataset. This is the model that we would like to parallelize across responses. |
progress |
A logical indicating whether to show a progress bar. |
f_multi A function that takes a formula and dataset and applies f to each response on the left hand side of the original formula.
mat <- data.frame(matrix(rnorm(100), 25, 4)) colnames(mat) <- c("y1", "y2", "x1", "x2") plm <- parallelize(lm) plm(y1 + y2 ~ x1 + x2, mat) prf <- parallelize(ranger::ranger) prf(mpg + hp ~ wt + disp + cyl, data = mtcars)mat <- data.frame(matrix(rnorm(100), 25, 4)) colnames(mat) <- c("y1", "y2", "x1", "x2") plm <- parallelize(lm) plm(y1 + y2 ~ x1 + x2, mat) prf <- parallelize(ranger::ranger) prf(mpg + hp ~ wt + disp + cyl, data = mtcars)
This is a helper function to visualize the raw data responsible for the largest indirect effects. It returns a faceted plot of outcome vs. mediator pairs for those with high pathwise indirect effects.
plot_mediators( indirect_effects, exper, n_digit = 3, n_panels = NULL, treatment = "treatment", ... )plot_mediators( indirect_effects, exper, n_digit = 3, n_panels = NULL, treatment = "treatment", ... )
indirect_effects |
A data.frame containing estimated indirect effects
for each variable, under different counterfactual settings for the "direct
treatment." This is the output of |
exper |
An object of class |
n_digit |
The number of digits of the indirect effects to print next to each panel. Defaults to 3. |
n_panels |
The number of mediator-outcome pairs to show. Defaults to 12, or the number of pathways, if there are fewer than 12. |
treatment |
What is the name of the treatment variable that we want to overlay on points? This is necessary when there are several potential treatment variables. Defaults to "treatment." |
... |
Further keyword arguments passed to |
A patchwork-based arrangement of ggplot2 grobs.
# dataset with no true effects exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) ie <- multimedia(exper) |> estimate(exper) |> indirect_pathwise() |> effect_summary() plot_mediators(ie, exper) # another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") ie <- multimedia(exper, rf_model()) |> estimate(exper) |> indirect_pathwise() |> effect_summary() plot_mediators(ie, exper)# dataset with no true effects exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) ie <- multimedia(exper) |> estimate(exper) |> indirect_pathwise() |> effect_summary() plot_mediators(ie, exper) # another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") ie <- multimedia(exper, rf_model()) |> estimate(exper) |> indirect_pathwise() |> effect_summary() plot_mediators(ie, exper)
This function draws a curve of indirect effect against the sensitivity parameter, allowing users to specify the name of x and y-axis variables using the x_var and y_var inputs.
plot_sensitivity(sensitivity_curve, x_var = "rho", y_var = "indirect_effect")plot_sensitivity(sensitivity_curve, x_var = "rho", y_var = "indirect_effect")
sensitivity_curve |
The output of a call to sensitivity or sensitivity_perturb. A data.frame whose columns are: outcome, {x_var}, {y_var}, and {y_var}_standard_error, where x_var and y_var are defined in the two arguments below. |
x_var |
The type of perturbation variable to plot along the x-axis. Defaults to "rho", following the sensitivity approach implemented in sensitivity_subset. |
y_var |
The type of effect to plot along the y-axis. Defaults to indirect_effect. |
A ggplot2 grob plotting the sensitivity parameter against the effect
specified by y_var.
sensitivity_curve <- read.csv(url("https://go.wisc.edu/j2kvcj")) plot_sensitivity(sensitivity_curve)sensitivity_curve <- read.csv(url("https://go.wisc.edu/j2kvcj")) plot_sensitivity(sensitivity_curve)
Predict across selected responses in a mediation model object. This is a lower-level version of the predict method that applies to objets of class mediation_model. Rather than giving predictions across all outcomes, it supports predictions across a subset specified as a vector. This can be convenient when we want to analyze how a specific subset of outcomes changes and do not need to run predictions across all possible.
predict_across(object, newdata, name)predict_across(object, newdata, name)
object |
An object of class |
newdata |
A data.frame containing new inputs from which to sample responses. If NULL, defaults to the data used to estimate fit. |
name |
A string or index specifying which of the dimensions of a multiresponse prediction to extract. |
A vector of predicted values for the outcome of interest.
exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) predict_across(outcome_model(fit), NULL, "outcome_1") # predict at newdata newdata <- bind_mediation(exper) predict_across( outcome_model(fit), newdata[seq_len(5), ], c("outcome_1", "outcome_2") )exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) predict_across(outcome_model(fit), NULL, "outcome_1") # predict at newdata newdata <- bind_mediation(exper) predict_across( outcome_model(fit), newdata[seq_len(5), ], c("outcome_1", "outcome_2") )
This generalizes the built-in predict method to the multimedia class. Given
an estimated multimedia object, this function supports prediction along the
estimated DAG. It first predicts hat[M] | T, X and then hat[Y] | hat[M], T, X. Each prediction step will call the prediction method within the
mediation and outcome models that make up the multimedia object on which this
is called. By passing in new treatment, mediator, or pretreatment data, you
can predict forward along the DAG using these as inputs.
## S4 method for signature 'multimedia' predict(object, profile = NULL, mediators = NULL, pretreatment = NULL, ...)## S4 method for signature 'multimedia' predict(object, profile = NULL, mediators = NULL, pretreatment = NULL, ...)
object |
An object of class multimedia containing the estimated mediation and outcome models whose mediation and outcome predictions we want to obtain. |
profile |
An object of class |
mediators |
By default, we will return outcome predictions using the
predicted mediators from the mediation model. Modify this argument if you
would like to directly control the mediation inputs for the outcome model.
Must be a data.frame whose columns are named to match the
|
pretreatment |
By default, we will return mediation and outcome model
predictions using the same pretreatment variables as used when initially
estimating the models (like setting |
... |
A placeholder to agree with |
A list with two elements: $mediators: A data.frame containing predicted values for the mediators. Each row corresponds to one row of the newdata, or one row of the default treatment profile, if no newdata is given.
$outcomes: A data.frame containing predicted values for the outcomes, given either (i) the predicted values of the mediators or (ii) the provided values of the mediators. Each row corresponds to one row of the newdata, or one row of the default treatment profile, if no newdata is given.
exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper, glmnet_model()) |> estimate(exper) predict(fit) # at new treatment configurations t1 <- data.frame(treatment = factor(rep(c(0, 1), each = 5))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t1) predict(fit, profile) # at new treatment and mediator configurations mediators <- data.frame(mediator = rnorm(10, 0, 1)) predict(fit, profile, mediators)exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper, glmnet_model()) |> estimate(exper) predict(fit) # at new treatment configurations t1 <- data.frame(treatment = factor(rep(c(0, 1), each = 5))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t1) predict(fit, profile) # at new treatment and mediator configurations mediators <- data.frame(mediator = rnorm(10, 0, 1)) predict(fit, profile, mediators)
This is an getter method for pretreatments in an S4 object, usually of class mediation_data.
pretreatments(object)pretreatments(object)
object |
An object whose pretreatments slot to modify. |
The output depends on the S4 class of the input object. If it is a multimedia model object, it will return a character vector of the outcome variable names. Otherwise it returns the outcome data.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) pretreatments(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) pretreatments(exper)
This is an accessor function to the @pretreatments slot in a mediation data object. It returns the entire set of observed pretreatments, in contrast to pretreatments() applied to a multimedia object, which only returns the names of the pretreatment variables.
## S4 method for signature 'mediation_data' pretreatments(object)## S4 method for signature 'mediation_data' pretreatments(object)
object |
An object of class mediation_data. |
A data.frame whose rows are samples and columns different pretreatments.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) pretreatments(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) pretreatments(exper)
Set Pretreatments This is an setter method for pretreatments in an S4 object, usually of class mediation_data.
pretreatments(object) <- valuepretreatments(object) <- value
object |
An object whose pretreatments slot to modify. |
value |
The new pretreatment values to set within object. |
Modifies the pretreatments slot of the input object in place.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) pretreatments(exper) <- data.frame(new_var = rnorm(100))exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) pretreatments(exper) <- data.frame(new_var = rnorm(100))
This is an setter method for the pretreatments slot in a mediation data object. It lets you supply a new pretreatments data.frame for the object.
## S4 replacement method for signature 'mediation_data' pretreatments(object) <- value## S4 replacement method for signature 'mediation_data' pretreatments(object) <- value
object |
An object of class mediation_data. |
value |
The new pretreatment values for the object. |
A version of object whose pretreatments slot has been replaced.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) pretreatments(exper) <- data.frame(x = 1:10) experexper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) pretreatments(exper) <- data.frame(x = 1:10) exper
This returns all the variables modeled within a multimedia object. This can be helpful for overviewing an experiment and is called by the print methods in this package. Also supports filtering to specific node types, e.g., mediators.
retrieve_names(object, nm)retrieve_names(object, nm)
object |
An object of class multimedia |
nm |
A string specifying the node type to filter down to, e.g., 'treatment' or 'mediator'. |
A character vector containing all the names of the variables of the
node type nm.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> retrieve_names("mediator")exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> retrieve_names("mediator")
Apply a random forest model in parallel across a vector of responses $y$ in either an outcome or mediation model. This is a natural choice when the relationship between inputs and outputs is thought to be nonlinear. Internally, each of the models across the response are estimated using the 'ranger' package.
rf_model(progress = TRUE, ...)rf_model(progress = TRUE, ...)
progress |
A logical indicating whether to show a progress bar during estimation. |
... |
Keyword parameters passed to ranger() in the 'ranger' package. |
model An object of class model with estimator, predictor, and
sampler functions associated wtih a lienar model.
model lm_model rf_model glmnet_model brms_model
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper, rf_model(num.trees = 10)) |> estimate(exper) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper, rf_model(num.trees = 20, max.depth = 2)) |> estimate(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper, rf_model(num.trees = 10)) |> estimate(exper) # example with another dataset exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") multimedia(exper, rf_model(num.trees = 20, max.depth = 2)) |> estimate(exper)
This assumes a continuous response, so that the out-of-sample MSE can be used
to estimate the outcome variability .
rf_sampler(fits, newdata = NULL, indices = NULL, ...)rf_sampler(fits, newdata = NULL, indices = NULL, ...)
fits |
The fitted RF model from which to draw samples. |
newdata |
A data.frame containing new inputs from which to sample responses. If NULL, defaults to the data used to estimate fit. |
indices |
The coordinates of the response from which to draw samples. |
... |
Additional parameters passed to rf_model's predict method. |
y_star A data.frame of samples y associated with the new inputs.
m <- rf_model() fit <- estimator(m)(mpg ~ hp + wt, data = mtcars) rf_sampler(fit, mtcars) prf <- parallelize(ranger::ranger) fit <- prf(mpg + disp ~ hp + wt, data = mtcars) rf_sampler(fit, mtcars)m <- rf_model() fit <- estimator(m)(mpg ~ hp + wt, data = mtcars) rf_sampler(fit, mtcars) prf <- parallelize(ranger::ranger) fit <- prf(mpg + disp ~ hp + wt, data = mtcars) rf_sampler(fit, mtcars)
This generalizes the built-in sample method to the multimedia class. Given an
estimated multimedia object, this function supports sampling along the
estimated DAG. It first samples M* | T, X and then Y* | M*, T, X. Each
sampling step will call the sample method within the mediation and outcome
models that make up the multimedia object on which this is called.
## S4 method for signature 'multimedia' sample(x, size, pretreatment = NULL, profile = NULL, mediators = NULL, ...)## S4 method for signature 'multimedia' sample(x, size, pretreatment = NULL, profile = NULL, mediators = NULL, ...)
x |
An object of class multimedia containing the estimated mediation and outcome models whose mediation and outcome samples we want to obtain. |
size |
A placeholder argument to agree with the default |
pretreatment |
By default, we will return mediation and outcome model
predictions using the same pretreatment variables as used when initially
estimating the models (like setting |
profile |
An object of class |
mediators |
By default, we will return outcome predictions using the
predicted mediators from the mediation model. Modify this argument if you
would like to directly control the mediation inputs for the outcome model.
Must be a data.frame whose columns are named to match the
|
... |
Additional options to pass to the @sampler method in the estimated mediation model. |
An object of class multimedia with mediator and outcome slots
sampled according to the description above.
exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) samples <- sample(fit) mediators(samples) outcomes(samples) # sampling with just different "n" has no effect. samples <- sample(fit, 100) # Instead sample at a new treatment configuration t1 <- data.frame(treatment = factor(rep(c(0, 1), each = 50))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t1) samples <- sample(fit, profile = profile) mediators(samples) outcomes(samples)exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) samples <- sample(fit) mediators(samples) outcomes(samples) # sampling with just different "n" has no effect. samples <- sample(fit, 100) # Instead sample at a new treatment configuration t1 <- data.frame(treatment = factor(rep(c(0, 1), each = 50))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t1) samples <- sample(fit, profile = profile) mediators(samples) outcomes(samples)
For causal identification, mediation analysis relies on several untestable assumptions. One important one is that there is no confounding between the counterfactual mediator and outcome variables. Even though we can never know whether this exists, we can measure the sensitivity of our conclusions to the existence/strength of such a confounder. In this function, we approach this by inducing (unallowable) correlation between the mediator and outcome model residuals, simulate forward, and see how the estimated overall indirect effect changes.
sensitivity( model, exper, confound_ix = NULL, rho_seq = NULL, n_bootstrap = 100, progress = TRUE, ... )sensitivity( model, exper, confound_ix = NULL, rho_seq = NULL, n_bootstrap = 100, progress = TRUE, ... )
model |
A |
exper |
The original |
confound_ix |
A data.frame specifying which mediator/outcome should be allowed to be correlated. Should have two columns: 'mediator' and 'outcome' specifying which pairs of mediators and outcomes should be correlated. Defaults to NULL, which creates a data.frame with no rows (and so enforcing independence between mediators and outcomes) |
rho_seq |
We will evaluate correlations Cor(e', e) between mediation and outcome model errors ranging along this grid. Defaults to NULL, which internally sets the sequence to rho = (-0.9, -0.7 ..., 0.7, 0.9). |
n_bootstrap |
The number of bootstrap resamples used to build confidence bands around the sensitivity curves. Defaults to 100. |
progress |
A logical indicating whether to show a progress bar. |
... |
Additional keyword arguments passed to |
A data.frame giving the outputs of indirect_overall across many
values of the correlation rho.
xy_data <- demo_spline() exper <- mediation_data( xy_data, starts_with("outcome"), "treatment", "mediator" ) model <- multimedia( exper, outcome_estimator = glmnet_model(lambda = 1e-2) ) |> estimate(exper) rho_seq <- c(-0.2, 0, 0.2) confound_ix <- expand.grid(mediator = 1, outcome = 1:2) sensitivity(model, exper, confound_ix, rho_seq, n_bootstrap = 2)xy_data <- demo_spline() exper <- mediation_data( xy_data, starts_with("outcome"), "treatment", "mediator" ) model <- multimedia( exper, outcome_estimator = glmnet_model(lambda = 1e-2) ) |> estimate(exper) rho_seq <- c(-0.2, 0, 0.2) confound_ix <- expand.grid(mediator = 1, outcome = 1:2) sensitivity(model, exper, confound_ix, rho_seq, n_bootstrap = 2)
For causal identification, mediation analysis relies on several untestable assumptions. One important one is that there is no confounding between the counterfactual mediator and outcome variables. Even though we can never know whether this exists, we can measure the sensitivity of our conclusions to the existence/strength of such a confounder. In this function, we approach this by inducing (unallowable) correlation between the mediator and outcome model residuals, simulate forward, and see how the estimated pathwise indirect effects change.
sensitivity_pathwise( model, exper, confound_ix = NULL, rho_seq = NULL, n_bootstrap = 100, progress = TRUE, ... )sensitivity_pathwise( model, exper, confound_ix = NULL, rho_seq = NULL, n_bootstrap = 100, progress = TRUE, ... )
model |
A |
exper |
The original |
confound_ix |
A data.frame specifying which mediator/outcome should be allowed to be correlated. Should have two columns: 'mediator' and 'outcome' specifying which pairs of mediators and outcomes should be correlated. Defaults to NULL, which creates a data.frame with no rows (and so enforcing independence between mediators and outcomes) |
rho_seq |
We will evaluate correlations Cor(e', e) between mediation and outcome model errors ranging along this grid. Defaults to NULL, which internally sets the sequence to rho = (-0.9, -0.7 ..., 0.7, 0.9). |
n_bootstrap |
The number of bootstrap resamples used to build confidence bands around the sensitivity curves. Defaults to 100. |
progress |
A logical indicating whether to show a progress bar. |
... |
Additional arguments passed to indirect_pathwise. |
A date.frame giving the outputs of indirect_overall across many
values of the correlation rho.
xy_data <- demo_spline() exper <- mediation_data( xy_data, starts_with("outcome"), "treatment", "mediator" ) model <- multimedia( exper, outcome_estimator = glmnet_model(lambda = 1e-2) ) |> estimate(exper) rho_seq <- c(-0.2, 0, 0.2) subset_indices <- expand.grid( mediator = n_mediators(model), outcome = n_outcomes(model) ) sensitivity_pathwise(model, exper, subset_indices, rho_seq, n_bootstrap = 2)xy_data <- demo_spline() exper <- mediation_data( xy_data, starts_with("outcome"), "treatment", "mediator" ) model <- multimedia( exper, outcome_estimator = glmnet_model(lambda = 1e-2) ) |> estimate(exper) rho_seq <- c(-0.2, 0, 0.2) subset_indices <- expand.grid( mediator = n_mediators(model), outcome = n_outcomes(model) ) sensitivity_pathwise(model, exper, subset_indices, rho_seq, n_bootstrap = 2)
The more standard sensitivity and sensitivity_pathwise functions support sensitivity analysis to violations in assumptions restricted to specific mediator-outcome pairs. For more general violations, this function allows arbitrary modification of the default, diagonal covariance matrix structure across both mediators and outcomes. This makes it possible to ask what happens when mediators are correlated with one another or when more some pairs of mediator-outcome pairs have much stronger correlation than others.
sensitivity_perturb( model, exper, perturb, nu_seq = NULL, n_bootstrap = 100, progress = TRUE )sensitivity_perturb( model, exper, perturb, nu_seq = NULL, n_bootstrap = 100, progress = TRUE )
model |
A |
exper |
The original |
perturb |
A matrix towards which the original mediator-outcome covariance should be perturbed. Must have dimension (n_mediators + n_outcomes) x (n_mediators + n_outcomes). |
nu_seq |
The strength of the perturbation towards the matrix perturb. |
n_bootstrap |
The number of bootstrap resamples used to build confidence bands around the sensitivity curves. Defaults to 100. |
progress |
A logical indicating whether to show a progress bar. |
Specifically, it defines a new covariance matrix across mediators and outcomes according to diag(sigma^2_mediator, sigma^2_outcome) + nu * perturb. The estimates sigma^2 are taken from the residuals in the original mediation and outcome models, and perturb and nu are provided by the user.
A date.frame giving the outputs of indirect_overall across many
values of the correlation rho.
xy_data <- demo_spline() exper <- mediation_data( xy_data, starts_with("outcome"), "treatment", "mediator" ) model <- multimedia( exper, outcome_estimator = glmnet_model(lambda = 1e-2) ) |> estimate(exper) nu_seq <- c(-0.2, 0.2) perturb <- matrix( c( 0, 3, 0, 3, 0, 0, 0, 0, 0 ), nrow = 3, byrow = TRUE ) sensitivity_perturb(model, exper, perturb, nu_seq, n_bootstrap = 2)xy_data <- demo_spline() exper <- mediation_data( xy_data, starts_with("outcome"), "treatment", "mediator" ) model <- multimedia( exper, outcome_estimator = glmnet_model(lambda = 1e-2) ) |> estimate(exper) nu_seq <- c(-0.2, 0.2) perturb <- matrix( c( 0, 3, 0, 3, 0, 0, 0, 0, 0 ), nrow = 3, byrow = TRUE ) sensitivity_perturb(model, exper, perturb, nu_seq, n_bootstrap = 2)
treatment_profile objectFor general mediation analysis, we need to provide counterfactuals for both
the outcome and mediator components of each sample. That is, we need to
understand Y(t, M(t')) where t and t' may not be the same.
treatment_profile classes place some more structural requirements on
treatment profiles, so that later effect estimation can make simplifying
assumptions. This function creates a treatment profile from a collection of
possible mediator and outcome treatments.
setup_profile(x, t_mediator = NULL, t_outcome = NULL)setup_profile(x, t_mediator = NULL, t_outcome = NULL)
x |
An object of class |
t_mediator |
A data.frame whose columns store treatment names and whose
values are the treatment assignments to each sample (row). Defaults to
NULL, in which case this type of data.frame is constructed from the
treatment assignments in the |
t_outcome |
A data.frame analogous to |
An object of class treatment_profile giving treatment assignments
for both mediation and outcome terms.
check_profile
exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) t1 <- data.frame(treatment = factor(rep(c(0, 1), each = 5))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t1) profile t2 <- data.frame(treatment = factor(rep(0, 10))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t2) profileexper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) t1 <- data.frame(treatment = factor(rep(c(0, 1), each = 5))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t1) profile t2 <- data.frame(treatment = factor(rep(0, 10))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t2) profile
This generates random spline functions. It is used in the toy nonlinear
dataset created in demo_spline(). This is not necessary for the essential
multimedia workflow, it is only exported for potential independent interest.
spline_fun(D = 2, knots = NULL, h_ix = seq_len(10), ...)spline_fun(D = 2, knots = NULL, h_ix = seq_len(10), ...)
D |
The number random spline functions to generate internally. |
knots |
The location of knots to use in the spline functions. Defaults to -4, -2, 0, 2, 4. |
h_ix |
The locations along which to generate the underlying spline function. |
... |
Additional arguments to pass to |
A function that can be used to sample points along the random spline functions. Takes argument 'x' giving the input to the spline and returns a D-dimensional response y giving random samples for each of the D splines..
demo_spline
Helper to Modify Formulas
sub_formula(formula, yj)sub_formula(formula, yj)
formula |
The original formula whose response we want to modify. |
yj |
The desired response term for the formula. |
A new formula object with the LHS replaced by yj.
This class ensures appropriate structure of the treatment assignments for
mediator and outcome variables. It enforces certain structural requirements
(e.g., that the number of samples is the same under the mediator and outcome
counterfactuals) using the check_profile function.
setup_profile check_profile
exper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) # helpers for defining treatment profiles t1 <- data.frame(treatment = factor(rep(c(0, 1), each = 5))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t1) profile t2 <- data.frame(treatment = factor(rep(0, 10))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t2) profileexper <- demo_spline(tau = c(2, 1)) |> mediation_data(starts_with("outcome"), "treatment", "mediator") fit <- multimedia(exper) |> estimate(exper) # helpers for defining treatment profiles t1 <- data.frame(treatment = factor(rep(c(0, 1), each = 5))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t1) profile t2 <- data.frame(treatment = factor(rep(0, 10))) profile <- setup_profile(fit, t_mediator = t1, t_outcome = t2) profile
This is an getter method for treatments in an S4 object, usually of class mediation_data.
treatments(object)treatments(object)
object |
An object whose treatments slot to modify. |
The output depends on the S4 class of the input object. If it is a multimedia model object, it will return a character vector of the outcome variable names. Otherwise it returns the outcome data.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) treatments(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) treatments(exper)
This is an accessor function to the @treatments slot in a mediation data object. It returns the entire set of observed treatments, in contrast to treatments() applied to a multimedia object, which only returns the names of the treatment variables.
## S4 method for signature 'mediation_data' treatments(object)## S4 method for signature 'mediation_data' treatments(object)
object |
An object of class mediation_data. |
A data.frame whose rows are samples and columns different treatments.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) treatments(exper)exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) treatments(exper)
This is a helper that returns the names of the treatment variables in an object of class multimedia. It parses the graph in the DAG specifying the mediation analysis, and it returns all variables of node type treatment.
## S4 method for signature 'multimedia' treatments(object)## S4 method for signature 'multimedia' treatments(object)
object |
An object of class multimedia. |
m A vector of strings containing the names of all the treatments.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> treatments()exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) multimedia(exper) |> treatments()
Set Treatments This is an setter method for treatments in an S4 object, usually of class mediation_data.
treatments(object) <- valuetreatments(object) <- value
object |
An object whose treatments slot to modify. |
value |
The new treatment values to set within object. |
Modifies the treatments slot of the input object in place.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) treatments(exper) <- data.frame(new_treatment = rnorm(100))exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) treatments(exper) <- data.frame(new_treatment = rnorm(100))
This is an setter method for the treatments slot in a mediation data object. It lets you supply a new treatments data.frame for the object.
## S4 replacement method for signature 'mediation_data' treatments(object) <- value## S4 replacement method for signature 'mediation_data' treatments(object) <- value
object |
An object of class mediation_data. |
value |
The new treatment values for the object. |
A version of object whose treatment slot has been replaced.
exper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) treatments(exper) <- data.frame(t = rep(0, 10)) experexper <- demo_joy() |> mediation_data("PHQ", "treatment", starts_with("ASV")) treatments(exper) <- data.frame(t = rep(0, 10)) exper