Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

https://github.com/moodymudskipper/ggframe

data frames that print as ggplots
https://github.com/moodymudskipper/ggframe

Last synced: 3 months ago
JSON representation

data frames that print as ggplots

Lists

README 

        
---

output: github_document

---



```{r, include = FALSE}

knitr::opts_chunk$set(

  collapse = TRUE,

  comment = "#>",

  fig.path = "man/figures/README-",

  out.width = "100%"

)

```



# ggframe



*{ggframe}* is a *{ggplot2}* wrapper. A "ggframe" object is a data.frame with

a "layers" and an "aes" attributes. It has a printing method which plots

by default, but print with `plot = FALSE` and you'll print the data, aesthetics

and layers.



You cannot do all you can do with *{ggplot2}*, it's just an exploration of some

ideas. Unlikely to go much further.



Some nice things that you can do more easily than in native *{ggplot2}* : 



* data wrangling on plot data (assuming you use functions that don't drop attributes, 

it seems *{dplyr}* functions don't, base R functions often do)

* keep track of the steps that built the plot.



## Installation



Install with:



``` r

remotes::install_github("moodymudskipper/ggframe")

```

## Functions



`map_aes()` sets an `aes` attribute.



`geom()`, `facet()`, `set_labs`, `set_coord()`, `set_theme()`, `set_scale()` 

and `set_guide()` are quick and dirty wrappers that

add the relevant plot building call to the `layer` attribute, 

I suppose a more serious version of the package would provide a wrapper for all

relevant *{ggplot2}* functions, with explicit arguments rather than just `...`. 

One advantage here is that it works with other packages extending *{ggplot2}*.



`as_ggplot()` converts a ggframe object to a standard ggplot object.



## Examples



```{r}

library(dplyr, warn.conflicts = FALSE)

library(ggplot2) # needs to be attached, geoms etc can come from other packages too

library(ggframe)



# fast to type for a quick plot 

iris %>%

  geom("point", aes(Petal.Length, Petal.Width, color = Species)) 



# defining default aes

iris %>%

  map_aes(Petal.Length, Petal.Width, color = Species) %>%

  geom("point") 



# wrangle the data after the layer definitions

iris %>%

  map_aes(Petal.Length, Petal.Width, color = Species) %>%

  geom("point") %>%

  filter(Species != "versicolor")



# print data, aesthetics and layers

iris %>%

  map_aes(Petal.Length, Petal.Width, color = Species) %>%

  geom("point") %>%

  slice_sample(n = 15) %>%

  print(plot = FALSE)



# use set_* functions to leverage other features

iris %>%

  map_aes(Petal.Length, Petal.Width, color = Species) %>%

  geom("point") %>%

  set_coord("fixed") %>%

  set_theme("classic") %>%

  set_scale("color_viridis_d") %>%

  set_labs(title = "hello")

```



In *{ggplot2}* the `ggplot()` functions takes as optional arguments the data and aesthetics

to be inherited by further layers. It doesn't link explicitly aesthetics to a datasets

(you can define aesthetics to be inherited, but use a different dataset for every layer) 

but it's quite rare to start without a default data layer nor default aesthetics.



Moreover when providing custom data for a layer, in my experience we generally don't

provide completely new data, but a transformation of the main data. Indeed the `data`

argument of geoms can be a function or formula to be applied on the main data.



It works quite nicely here:



```{r}

iris %>%

  map_aes(Petal.Length, Petal.Width) %>%

  geom("point", data = head, color = "blue") %>%

  geom("point", data = tail, color = "red")

```



Thanks to the fact that `ggplot2::ggplot_build()` is generic and called by

`layer_data()`, `layer_scales()` and `layer_grob()`, we could make those

work directly on ggframe objects :



```{r}

# leverage layer_data easily

iris %>%

  map_aes(Petal.Length, Petal.Width) %>%

  geom("point", data = head, color = "blue") %>%

  geom("point", data = tail, color = "red") %>%

  layer_data(2)

```