README.Rmd

---
output: github_document
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)
```
# tidyroll

<!-- badges: start -->
[![Lifecycle: experimental](https://img.shields.io/badge/lifecycle-experimental-orange.svg)](https://www.tidyverse.org/lifecycle/#experimental)
[![Codecov test coverage](https://codecov.io/gh/gacolitti/tidyroll/branch/master/graph/badge.svg)](https://codecov.io/gh/gacolitti/tidyroll?branch=master)
[![R build status](https://github.com/gacolitti/tidyroll/workflows/R-CMD-check/badge.svg)](https://github.com/gacolitti/tidyroll/actions)
<!-- badges: end -->

`tidyroll` makes it easy to work with irregular time slices for modeling and prediciton with `tidymodels`.

The main function `rolling_origin_nested()` is a wrapper around `rsample::rolling_origin` and facilitates
rolling over different time units instead of a fixed window. The motivation for this function comes
from this [vignette](https://tidymodels.github.io/rsample/articles/Applications/Time_Series.html). 

## Installation

```{r, eval = FALSE}
# install.packages("devtools")
devtools::install_github("gacolitti/tidyroll")
```

## Example

```{r example}
suppressPackageStartupMessages(library(tidyroll))
suppressPackageStartupMessages(library(dplyr))
suppressPackageStartupMessages(library(rsample))

data("airquality2")

roll <- rolling_origin_nested(
  data = airquality2, 
  time_var = "date", 
  unit = "month", 
  start = "2017-08-01",
  end = "2017-11-01",
  assess = 1
)
roll

analysis(roll$splits[[1]]) 
assessment(roll$splits[[1]]) 

analysis(roll$splits[[1]])$data %>% last() %>% tail()
assessment(roll$splits[[1]])$data %>% first() %>% head()
```