https://github.com/poissonconsulting/newdata
An R Package to Generate New Data Frames for Prediction
https://github.com/poissonconsulting/newdata
Last synced: about 1 month ago
JSON representation
An R Package to Generate New Data Frames for Prediction
- Host: GitHub
- URL: https://github.com/poissonconsulting/newdata
- Owner: poissonconsulting
- License: other
- Created: 2016-09-15T13:37:55.000Z (about 9 years ago)
- Default Branch: main
- Last Pushed: 2025-07-31T23:50:57.000Z (2 months ago)
- Last Synced: 2025-08-01T01:38:41.741Z (2 months ago)
- Language: R
- Homepage: https://poissonconsulting.github.io/newdata/
- Size: 8.44 MB
- Stars: 4
- Watchers: 3
- Forks: 1
- Open Issues: 3
-
Metadata Files:
- Readme: README.Rmd
- Changelog: NEWS.md
- Contributing: .github/CONTRIBUTING.md
- License: LICENSE
- Code of conduct: .github/CODE_OF_CONDUCT.md
- Support: .github/SUPPORT.md
Awesome Lists containing this project
- jimsghstars - poissonconsulting/newdata - An R Package to Generate New Data Frames for Prediction (R)
README
---
output: github_document
---
```{r, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
```
# newdata 
[](https://lifecycle.r-lib.org/articles/stages.html#stable)
[](https://github.com/poissonconsulting/newdata/actions/workflows/R-CMD-check.yaml)
[](https://codecov.io/gh/poissonconsulting/newdata)
[](https://opensource.org/license/mit)
[](https://cran.r-project.org/package=newdata)
## Introduction
`newdata` is an R package to generate new data frames by varying some variables while holding the others constant.
By default, all specified variables vary across their range
while all other variables are held constant at a reference value.
The user can specify the length of each sequence, require that only
observed values and combinations are used and add new variables.
Types, classes, factor levels and time zones are always preserved.
Consider the following observed 'old' data frame.
```{r}
library(newdata)
newdata::old_data
```
### Reference Value
By default all variables are set to a reference value.
```{r}
xnew_data(old_data)
```
The reference value depends on the class of the variable, by default:
- logical vectors are FALSE;
- double vectors are the mean;
- integer, Date, POSIXct and hms vectors are the floored mean;
- character vectors are the most common value or the first when sorted of the most common values;
- factor and ordered vectors are the first level.
### Sequences
Specifying a variable causes it to vary sequentially across its range.
```{r}
xnew_data(old_data, int)
```
By default the sequence depends on the class of the variable:
- logical vectors are length 2 (TRUE and FALSE);
- double vectors are 30 equally spaced values from the minimum value to the maximum value;
- integer, Date, POSIXct and hms vectors are up to 30 discrete values from the minimum to the maximum value as evenly spaced as possible;
- character vectors are the number of unique values.
- factor and ordered vectors are the number of levels.
These values can be overridden by setting the following options:
- `new_data.length_out_lgl`, which is 2 by default, for logical vectors;
- `new_data.length_out_dbl`, which is 30 by default, for double vectors;
- `new_data.length_out_int`, which is 30 by default, for integer, Date, POSIXct and hms vectors^1;
- `new_data.length_out_chr`, which is Inf by default, for character, factor and ordered vectors.
1. The length of Date, POSIXct and hms sequences are controlled by `new_data.length_out_int` as they are treated as integers for the purpose of generating a sequence.
When programming it is strongly recommended that the user explicitly specify the length of each sequence individually.
```{r}
xnew_data(old_data, lgl, xnew_seq(int, length_out = 3))
```
A third alternative is to specify the length of all the sequences in the data set but this can result in less common character strings or later factor or ordered levels being dropped.
```{r}
xnew_data(old_data, dbl, int, .length_out = 2)
```
### Observed Values
The user can also indicate whether only observed values should be used in the sequence.
```{r}
xnew_data(old_data, xnew_seq(int, length_out = 3, obs_only = TRUE))
```
The `xobs_only()` function can be used to filter out unobserved values after the sequence has been generated.
```{r}
xnew_data(old_data, xobs_only(xnew_seq(int, length_out = 3)))
```
and when two or more variables are specified all combinations are used.
```{r}
xnew_data(old_data, int, fct)
```
to only get observed combinations.
```{r}
xnew_data(old_data, xobs_only(int, fct))
```
### Modifying Variables
Modifying an existing variable or changing an existing one is simple.
```{r}
xnew_data(old_data, lgl = median(lgl, na.rm = TRUE), extra = c(TRUE, FALSE))
```
### Casting Variables
Casting variables to be the same class as the original is achieved as follows.
```{r}
xnew_data(old_data, xcast(lgl = 1, int = 7, dbl = 10L, fct = "a rarity", hms = "00:00:02"))
```
### A Simple Wrapper
Although superseded, for consistency with existing code `new_data()` which is a simple wrapper on `xnew_data()` allows the user to pass a character vector and to specifying the length of all the sequences is also provided.
```{r}
new_data(old_data, seq = c("int", "fct"), length_out = 5)
```
## Installation
To install the latest release version from CRAN.
```r
install.packages("newdata")
```
To install the latest development version from [GitHub](https://github.com/poissonconsulting/newdata)
```r
# install.packages("pak")
pak::pak("poissonconsulting/newdata")
```
or from [r-universe](https://poissonconsulting.r-universe.dev/newdata).
```r
install.packages("newdata", repos = c("https://poissonconsulting.r-universe.dev", "https://cloud.r-project.org"))
```
## Contribution
Please report any [issues](https://github.com/poissonconsulting/newdata/issues).
[Pull requests](https://github.com/poissonconsulting/newdata/pulls) are always welcome.
## Code of Conduct
Please note that the newdata project is released with a [Contributor Code of Conduct](https://contributor-covenant.org/version/2/1/CODE_OF_CONDUCT.html). By contributing to this project, you agree to abide by its terms.