---
title: "Leave-one-out D_i diagnostics"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Leave-one-out D_i diagnostics}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>")
set.seed(20260607)
```

The function `compute_Di()` computes fixed-grid leave-one-out self-support
scores:

\[
D_i = 1 - \frac{\widehat f_{(-i)}(X_i)}{\widehat f(X_i)}.
\]

The bandwidths, grid origin and estimator parameters are held fixed. Only the
contribution of observation `i` is removed. The result is a diagnostic for the
chosen estimator and grid, not a standalone model-selection criterion.

```{r}
library(GLBFP)

x <- cbind(rnorm(120), rnorm(120))
b <- c(0.7, 0.7)
m <- c(2, 2)

scores <- compute_Di(x, b = b, m = m, estimator = "GLBFP")
scores
summary(scores)
```

The output can be converted to a data frame.

```{r}
score_tbl <- as.data.frame(scores)
head(score_tbl)
```

The S3 plot method supports an index plot and a density-versus-score plot.

```{r}
plot(scores)
plot(scores, type = "density")
```

The same interface supports `LBFP` and `ASH`.

```{r}
lbfp_scores <- compute_di(x, b = b, estimator = "LBFP")
ash_scores <- compute_di(x, b = b, m = m, estimator = "ASH")

c(
  LBFP_mean = mean(lbfp_scores$D),
  ASH_mean = mean(ash_scores$D),
  GLBFP_mean = mean(scores$D)
)
```

Interpretation depends on the chosen estimator, bandwidth and grid. Large
positive values indicate observations whose removal substantially decreases
their own fitted support under the fixed-grid estimator.