Fit a multimodal response model — fit_gazepoint_multimodal_response

Fits an explicit response model using multimodal predictors. Linear models, generalised linear models, and mixed-effects models are supported depending on family and random_effects. The helper prepares and fits the model only; it does not make causal, diagnostic, or emotion-inference claims.

Usage

fit_gazepoint_multimodal_response_model(
  data,
  outcome,
  predictors,
  covariates = NULL,
  random_effects = NULL,
  family = NULL,
  na_action = c("na.omit", "na.exclude"),
  REML = FALSE,
  ...
)

Arguments

data: A multimodal analysis data frame, usually returned by prepare_gazepoint_multimodal_data().
outcome: Outcome column name.
predictors: Character vector of fixed-effect predictor columns.
covariates: Optional character vector of covariate columns.
random_effects: Optional random-effects formula component, for example "(1 | participant_id)".
family: Optional model family. If NULL, stats::lm() or lme4::lmer() is used. If supplied, stats::glm() or lme4::glmer() is used.
na_action: Missing-data handling. One of "na.omit" or "na.exclude".
REML: Passed to lme4::lmer() when a linear mixed model is fitted.
...: Additional arguments passed to the model-fitting function.

Value

A list with model, formula, data, and settings. The object has class gp3_multimodal_response_model.

Examples

dat <- data.frame(
  participant_id = c("P001", "P002", "P003"),
  AU12_r_mean = c(0.1, 0.2, 0.3),
  dwell_time = c(1.2, 1.4, 1.1),
  rating = c(3, 4, 5)
)

fit_gazepoint_multimodal_response_model(
  dat,
  outcome = "rating",
  predictors = c("AU12_r_mean", "dwell_time")
)
#> $model
#> 
#> Call:
#> stats::lm(formula = form, data = analysis_data, na.action = na_fun)
#> 
#> Coefficients:
#> (Intercept)  AU12_r_mean   dwell_time  
#>    2.00e+00     1.00e+01    -2.93e-15  
#> 
#> 
#> $formula
#> rating ~ AU12_r_mean + dwell_time
#> <environment: 0x55afe5be7ac0>
#> 
#> $data
#> # A tibble: 3 × 4
#>   participant_id AU12_r_mean dwell_time rating
#>   <chr>                <dbl>      <dbl>  <dbl>
#> 1 P001                   0.1        1.2      3
#> 2 P002                   0.2        1.4      4
#> 3 P003                   0.3        1.1      5
#> 
#> $settings
#> $settings$model_label
#> [1] "multimodal_response_model"
#> 
#> $settings$outcome
#> [1] "rating"
#> 
#> $settings$predictors
#> [1] "AU12_r_mean" "dwell_time" 
#> 
#> $settings$covariates
#> NULL
#> 
#> $settings$random_effects
#> NULL
#> 
#> $settings$family
#> NULL
#> 
#> $settings$na_action
#> [1] "na.omit"
#> 
#> $settings$REML
#> [1] FALSE
#> 
#> $settings$n_rows_input
#> [1] 3
#> 
#> $settings$n_rows_model
#> [1] 3
#> 
#> 
#> attr(,"class")
#> [1] "gp3_multimodal_response_model" "gp3_multimodal_model"         
#> [3] "list"