Create directional heatmaps across continuous & binary variables — PlotDirectionalHeatmaps • SciDataReportR

Combines:

Continuous~Continuous (Pearson/Spearman)
Binary~Binary (Phi; 1 == PositiveLevel)
Binary~Continuous (r_pb; 1 == PositiveLevel) into a single square heatmap with raw and FDR-star overlays.

Usage

PlotDirectionalHeatmaps(
  data,
  variables = NULL,
  Relabel = TRUE,
  Ordinal = TRUE,
  fdr_scope = c("matrix", "per_outcome", "per_predictor"),
  Data = lifecycle::deprecated(),
  xVars = lifecycle::deprecated(),
  yVars = lifecycle::deprecated()
)

Arguments

data: A dataframe.
variables: Character vector of variables to include (subset of data columns). The analysis is symmetric: every variable is related to every other, so a single variable set defines both axes. If NULL, uses all detected continuous + binary vars.
Relabel: Logical; use sjlabelled variable labels if present.
Ordinal: Logical; passed to PlotPointCorrelationsHeatmap() for the binary~continuous block, where it controls whether ordinal variables are treated as continuous. Defaults to TRUE.
fdr_scope: Either "matrix" (default) or "per_outcome", threaded through to the three sub-analyses (PlotCorrelationsHeatmap(), PlotPhiHeatmap(), PlotPointCorrelationsHeatmap()). Correction is applied within each sub-analysis block (continuous~continuous, binary~binary, binary~continuous), matching historical behavior; each sub-function's documented outcome orientation applies within its block.
Data: Deprecated (since 19.15.0). Use data instead.
xVars: Deprecated (since 19.15.0). Use variables instead.
yVars: Deprecated (since 19.15.0). Use variables instead. If supplied, the old rectangular x-by-y display is still honored.

Value

list(Unadjusted, FDRCorrected, Relabel, BinaryMapping, Excluded)

Details

Constant variables (no variation in the current data) are automatically excluded before computing any tiles.

Note

The analysis covers continuous and binary variables. Multi-level categorical variables (more than two levels) are not placed on the heatmap. Ordinal affects only the binary~continuous block, which is the one sub-analysis that accepts it.

Examples

data(SampleData)
data(SampleVariableTypes)

# Attach labels and factor levels so binary variables are detected and
# axis labels are readable
Labelled <- RevalueData(SampleData, SampleVariableTypes)$RevaluedData

# A mix of binary categorical (Diagnosis, sex) and continuous variables
result <- PlotDirectionalHeatmaps(
  Labelled,
  variables = c("Diagnosis", "sex", "age", "AXL", "Adiponectin",
                "Alpha_1_Antitrypsin", "C_Reactive_Protein", "Cortisol",
                "Insulin", "Leptin")
)

# Raw p-value directional heatmap
result$Unadjusted$plot
#> Warning: Removed 24 rows containing missing values or values outside the scale range
#> (`geom_text()`).


# FDR-adjusted directional heatmap
result$FDRCorrected$plot
#> Warning: Removed 26 rows containing missing values or values outside the scale range
#> (`geom_text()`).


# How binary variables were coded (which level counts as the positive one)
result$BinaryMapping
#>    Variable     Label PositiveLevel NegativeLevel
#> 1 Diagnosis Diagnosis      Impaired       Control
#> 2       sex       Sex          Male        Female