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

```{r setup, include = FALSE}
library(tidypredict)
library(dplyr)
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

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

`parsnip::nullmodel()` fits a model that ignores the predictors entirely. For
regression it predicts the mean of the outcome, and for classification it
predicts the observed class frequencies. Predictions are therefore constants, and
`tidypredict_fit()` returns a plain number for regression and a *named list* of
one constant per class for classification. Since the classification output is a
list, `tidypredict_to_column()` and `tidypredict_test()` are only supported for
regression.

## `tidypredict_` functions

```{r}
model <- parsnip::nullmodel(mtcars[-1], mtcars$mpg)
```

- 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)
    ```

For classification, one expression per class is returned:

```{r}
c_model <- parsnip::nullmodel(iris[-5], iris$Species)

tidypredict_fit(c_model)
```

## parsnip

`parsnip` fitted models are also supported by `tidypredict`:
```{r}
library(parsnip)

p_model <- null_model(mode = "regression") %>%
  set_engine("parsnip") %>%
  fit(mpg ~ ., data = mtcars)
```

```{r}
tidypredict_fit(p_model)
```

## Parse model spec

Here is an example of the model spec:
```{r}
pm <- parse_model(model)
str(pm, 2)
```