---
title: "Package overview and workflow map"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Package overview and workflow map}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>")
set.seed(2026)
```

This vignette gives a compact map of the package. For a first executable
example, start with the "Getting started with GLBFP" vignette.

## Estimator families

`GLBFP` implements three related histogram-based density estimator families.

| Family | Pointwise function | Grid function | Main tuning inputs |
|---|---|---|---|
| Averaged Shifted Histogram | `ASH()` | `ASH_estimate()` | `b`, `m` |
| Linear Blend Frequency Polygon | `LBFP()` | `LBFP_estimate()` | `b` |
| General Linear Blend Frequency Polygon | `GLBFP()` | `GLBFP_estimate()` | `b`, `m` |

Lowercase aliases are also available:

```{r}
library(GLBFP)

c(
  ash = exists("ash"),
  lbfp = exists("lbfp"),
  glbfp = exists("glbfp")
)
```

The uppercase names are kept for compatibility with the original package API.

## Core workflow

Most analyses follow the same sequence:

1. prepare a finite numeric matrix or data frame;
2. choose a bandwidth vector `b`;
3. choose `m` when using `ASH` or `GLBFP`;
4. evaluate a pointwise estimator or a grid estimator;
5. inspect the S3 summary, plot, and optional diagnostics.

```{r}
x <- matrix(rnorm(300), ncol = 1)
b <- compute_bi_optim(x, m = 1)

point_fit <- glbfp(x = 0, data = x, b = b, m = 1)
grid_fit <- glbfp_estimate(data = x, b = b, m = 1, grid_size = 80)

summary(point_fit)
summary(grid_fit)
```

## Article guide

Use the vignettes in this order when learning the package:

| Article | Purpose |
|---|---|
| Getting started with GLBFP | First runnable examples and basic object usage |
| Package overview and workflow map | Orientation across estimators and workflows |
| Brief methodological background | Minimal statistical context and references |
| Choosing between ASH, LBFP and GLBFP | Practical estimator comparison |
| Two-dimensional density estimation | 2D estimation and visualization |
| Sparse-prefix computation | Internal sparse grid-count diagnostics |
| Leave-one-out `D_i` diagnostics | Fixed-grid leave-one-out diagnostics |
| Objects, summaries and plotting | S3 classes and helper methods |
| Validation and comparison | Lightweight benchmark scaffolding |

## Input expectations

The package expects finite numeric observations with one observation per row.
Missing values should be removed or handled before estimation. For nearly
constant data, provide explicit non-degenerate `min_vals` and `max_vals`.