---
title: "bagger models"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{bagger models}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include = FALSE}
if (requireNamespace("baguette", quietly = TRUE)) {
  library(tidypredict)
  library(dplyr)
  eval_code <- TRUE
} else {
  eval_code <- FALSE
}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  eval = eval_code
)
```

| Function                                                      |Works|
|---------------------------------------------------------------|-----|
|`tidypredict_fit()`, `tidypredict_sql()`, `parse_model()`      |  ✔  |
|`tidypredict_to_column()`                                      |  ✔  |
|`tidypredict_test()`                                           |  ✔  |
|`tidypredict_interval()`, `tidypredict_sql_interval()`         |  ✗  |
|`parsnip`                                                      |  ✔  |

`baguette::bagger()` fits an ensemble of models on bootstrap samples of the
training data. The `"CART"` base model, which fits `rpart::rpart()` trees, and
the `"C5.0"` base model, which fits `C50::C5.0()` trees, are supported.
`tidypredict_fit()` returns one nested `case_when()` per tree, so the size of
the returned expression grows with `times`.

For regression models the fitted value is the mean of the individual tree
predictions. For classification models the class probabilities of each tree are
averaged, and the returned expression is the class with the largest average
probability.

## `tidypredict_` functions

```{r}
set.seed(100)
model <- baguette::bagger(mpg ~ wt + cyl + disp, data = mtcars, times = 5)
```

- Create the R formula
    ```{r}
tidypredict_fit(model)
    ```

- Add the predictions to the original table
    ```{r}
mtcars %>%
  tidypredict_to_column(model) %>%
  glimpse()
    ```

- Confirm that the results match the model's `predict()` results
    ```{r}
tidypredict_test(model, mtcars)
    ```

- Get the SQL translation
    ```{r}
tidypredict_sql(model, dbplyr::simulate_mssql())
    ```

## Classification

```{r}
set.seed(100)
model <- baguette::bagger(Species ~ ., data = iris, times = 3)

tidypredict_test(model, iris)
```

C5.0 trees are only fit for classification, and are used by passing
`base_model = "C5.0"`.

```{r}
set.seed(100)
model <- baguette::bagger(
  Species ~ .,
  data = iris,
  base_model = "C5.0",
  times = 3
)

tidypredict_test(model, iris)
```

## parsnip

Models fit with `parsnip::bag_tree()` and the `"rpart"` or `"C5.0"` engine are
supported as well.

```{r}
library(parsnip)

set.seed(100)
model <- bag_tree(mode = "regression") %>%
  set_engine("rpart", times = 5) %>%
  fit(mpg ~ wt + cyl + disp, data = mtcars)

tidypredict_fit(model)
```