Impute missing values in Gazepoint signals

Interpolates missing values in numeric Gazepoint time series, such as pupil, GSR/EDA, PPG/BVP, heart-rate, IBI/RRI, or other continuous channels. The function can work on a numeric vector, a time-series object, or selected numeric columns of a data frame.

Usage

impute_gazepoint_missing(
  data,
  method = c("linear", "locf", "nocb", "nearest", "constant"),
  cols = NULL,
  time_col = NULL,
  group_cols = NULL,
  max_gap = Inf,
  fill_edges = TRUE,
  constant_value = 0,
  add_flags = TRUE,
  treat_infinite_as_missing = TRUE
)

Arguments

data

Numeric vector, time-series object, or data frame.

method

Imputation method: "linear", "locf", "nocb", "nearest", or "constant".

cols

Columns to impute when data is a data frame. If NULL, all numeric columns except time and grouping columns are used.

time_col

Optional time column for interpolation.

group_cols

Optional grouping columns. Imputation is performed within groups.

max_gap

Maximum missing-gap length, in samples, to impute. Longer gaps remain missing. Defaults to Inf.

fill_edges

If TRUE, leading and trailing gaps are filled using the nearest observed value for methods that support it.

constant_value

Value used when method = "constant".

add_flags

If TRUE and data is a data frame, add logical <column>_was_imputed columns.

treat_infinite_as_missing

If TRUE, infinite values are treated as missing before imputation.

Value

Object of the same basic type as data. Data-frame outputs include an imputation_summary attribute.

Examples

x <- c(1, NA, 3, 4)
impute_gazepoint_missing(x)
#> [1] 1 2 3 4

dat <- data.frame(time_s = 1:5, GSR = c(1, NA, 3, NA, 5))
impute_gazepoint_missing(dat, cols = "GSR", time_col = "time_s")
#>   time_s GSR GSR_was_imputed
#> 1      1   1           FALSE
#> 2      2   2            TRUE
#> 3      3   3           FALSE
#> 4      4   4            TRUE
#> 5      5   5           FALSE