Introduction to teal.data

Introduction

The teal.data package specifies the data format used in teal applications. The teal_data class inherits from qenv and is meant to be used for reproducibility purposes.

Quick Start

To create an object of class teal_data, use the teal_data function. teal_data has a number of methods to manage relevant information in private class slots.

library(teal.data)

# create teal_data object
my_data <- teal_data()

# run code within teal_data to create data objects
my_data <- within(
  my_data,
  {
    data1 <- data.frame(id = 1:10, x = 11:20)
    data2 <- data.frame(id = 1:10, x = 21:30)
    data3 <- data.frame(id = 1:10, x = 31:40)
  }
)

# get objects stored in teal_data
my_data[["data1"]]
my_data[["data2"]]

# limit objects stored in teal_data
my_data[c("data1", "data3")]

# get reproducible code
get_code(my_data)

# get code just for specific object
get_code(my_data, names = "data2")
# get datanames
names(my_data)

# print
print(my_data)

teal_data characteristics

A teal_data object keeps the following information:

  • env - an environment containing data.
  • code - a string containing code to reproduce env (details in reproducibility).
  • names - a character vector listing objects of interest to teal modules (details in this teal vignette).
  • join_keys - a join_keys object defining relationships between datasets (details in Join Keys).

Reproducibility

The primary function of teal_data is to provide reproducibility of data. We recommend to initialize empty teal_data, which marks object as verified, and create datasets by evaluating code in the object, using within or eval_code. Read more in teal_data Reproducibility.

my_data <- teal_data()
my_data <- within(my_data, data <- data.frame(x = 11:20))
my_data <- within(my_data, data$id <- seq_len(nrow(data)))
my_data # is verified
## ✅︎ verified teal_data object
## <environment: 0x55bd3196a980> [L]
## Parent: <environment: package:teal.data>
## Bindings:
## • data: <data.frame> [L]

Relational data models

The teal_data class supports relational data. Relationships between datasets can be described by joining keys and stored in a teal_data object. These relationships can be read or set with the join_keys function. See more in join_keys.

my_data <- teal_data()
my_data <- within(my_data, {
  data <- data.frame(id = 1:10, x = 11:20)
  child <- data.frame(id = 1:20, data_id = c(1:10, 1:10), y = 21:30)
})

join_keys(my_data) <- join_keys(
  join_key("data", "data", key = "id"),
  join_key("child", "child", key = "id"),
  join_key("child", "data", key = c("data_id" = "id"))
)

join_keys(my_data)
## A join_keys object containing foreign keys between 2 datasets:
## child: [id]
##   <-- data: [id]
## data: [id]
##   --> child: [data_id]
# join_keys for limited object
join_keys(my_data["child"])
## A join_keys object containing foreign keys between 1 datasets:
## child: [id]

Hidden objects

An object is hidden in teal_data if its name starts with a dot (.). This can be used to pass auxiliary objects / function in the teal_data instance, without being visible in the teal summary and filter panel.

my_data <- teal_data()
my_data <- within(my_data, {
  data <- data.frame(id = 1:10, x = 11:20)
  .data2 <- data.frame(id = 1:20, data_id = c(1:10, 1:10), y = 21:30)
})

names(my_data)
## [1] "data"