Title: Data Base Backend for 'mlr3'
Version: 0.6.0
Description: Extends the 'mlr3' package with a backend to transparently work with databases such as 'SQLite', 'DuckDB', 'MySQL', 'MariaDB', or 'PostgreSQL'. The package provides three additional backends: 'DataBackendDplyr' relies on the abstraction of package 'dbplyr' to interact with most DBMS. 'DataBackendDuckDB' operates on 'DuckDB' data bases and also on Apache Parquet files. 'DataBackendPolars' operates on 'Polars' data frames.
License: LGPL-3
URL: https:///mlr3db.mlr-org.com, https://github.com/mlr-org/mlr3db
BugReports: https://github.com/mlr-org/mlr3db/issues
Depends: mlr3 (≥ 0.13.0), R (≥ 3.1.0)
Imports: backports, checkmate, data.table, mlr3misc (≥ 0.10.0), R6
Suggests: DBI, dbplyr, dplyr, duckdb (≥ 0.4.0), future, future.apply, future.callr, lgr, polars, RSQLite, testthat (≥ 3.0.0), tibble
Additional_repositories: https://community.r-multiverse.org
Config/testthat/edition: 3
Encoding: UTF-8
RoxygenNote: 7.3.2
NeedsCompilation: no
Packaged: 2025-07-17 11:41:01 UTC; marc
Author: Michel Lang ORCID iD [aut], Lona Koers [aut], Marc Becker ORCID iD [cre, aut]
Maintainer: Marc Becker <marcbecker@posteo.de>
Repository: CRAN
Date/Publication: 2025-07-18 08:40:02 UTC

mlr3db: Data Base Backend for 'mlr3'

Description

Extends the 'mlr3' package with a backend to transparently work with databases such as 'SQLite', 'DuckDB', 'MySQL', 'MariaDB', or 'PostgreSQL'. The package provides two additional backends: 'DataBackendDplyr' relies on the abstraction of package 'dbplyr' to interact with most DBMS. 'DataBackendDuckDB' operates on 'DuckDB' data bases and also on Apache Parquet files.

Options

Author(s)

Maintainer: Michel Lang michellang@gmail.com (ORCID)

Authors:

See Also

Useful links:

DataBackend for dplyr/dbplyr

Description

A mlr3::DataBackend using dplyr::tbl() from packages dplyr/dbplyr. This includes tibbles and abstract database connections interfaced by dbplyr. The latter allows mlr3::Tasks to interface an out-of-memory database.

Super class

mlr3::DataBackend -> DataBackendDplyr

Public fields

levels

(named list())
List (named with column names) of factor levels as character(). Used to auto-convert character columns to factor variables.

connector

(⁠function()⁠)
Function which is called to re-connect in case the connection became invalid.

Active bindings

rownames

(integer())
Returns vector of all distinct row identifiers, i.e. the contents of the primary key column.

colnames

(character())
Returns vector of all column names, including the primary key column.

nrow

(integer(1))
Number of rows (observations).

ncol

(integer(1))
Number of columns (variables), including the primary key column.

valid

(logical(1))
Returns NA if the data does not inherits from "tbl_sql" (i.e., it is not a real SQL data base). Returns the result of DBI::dbIsValid() otherwise.

Methods

Public methods

Inherited methods

Method new()

Creates a backend for a dplyr::tbl() object.

Usage
DataBackendDplyr$new(
  data,
  primary_key,
  strings_as_factors = TRUE,
  connector = NULL
)
Arguments
data

(dplyr::tbl())
The data object.

Instead of calling the constructor yourself, you can call mlr3::as_data_backend() on a dplyr::tbl(). Note that only objects of class "tbl_lazy" will be converted to a DataBackendDplyr (this includes all connectors from dbplyr). Local "tbl" objects such as tibbles will converted to a mlr3::DataBackendDataTable.

primary_key

(character(1))
Name of the primary key column.

strings_as_factors

(logical(1) || character())
Either a character vector of column names to convert to factors, or a single logical flag: if FALSE, no column will be converted, if TRUE all string columns (except the primary key). For conversion, the backend is queried for distinct values of the respective columns on construction and their levels are stored in ⁠$levels⁠.

connector

(function()⁠)\cr If not ⁠NULL', a function which re-connects to the database in case the connection has become invalid. Database connections can become invalid due to timeouts or if the backend is serialized to the file system and then de-serialized again. This round trip is often performed for parallelization, e.g. to send the objects to remote workers. DBI::dbIsValid() is called to validate the connection. The function must return just the connection, not a dplyr::tbl() object! Note that this this function is serialized together with the backend, including possible sensitive information such as login credentials. These can be retrieved from the stored mlr3::DataBackend/mlr3::Task. To protect your credentials, it is recommended to use the secret package.

Method data()

Returns a slice of the data. Calls dplyr::filter() and dplyr::select() on the table and converts it to a data.table::data.table().

The rows must be addressed as vector of primary key values, columns must be referred to via column names. Queries for rows with no matching row id and queries for columns with no matching column name are silently ignored. Rows are guaranteed to be returned in the same order as rows, columns may be returned in an arbitrary order. Duplicated row ids result in duplicated rows, duplicated column names lead to an exception.

Usage
DataBackendDplyr$data(rows, cols)
Arguments
rows

integer()
Row indices.

cols

character()
Column names.

Method head()

Retrieve the first n rows.

Usage
DataBackendDplyr$head(n = 6L)
Arguments
n

(integer(1))
Number of rows.

Returns

data.table::data.table() of the first n rows.

Method distinct()

Returns a named list of vectors of distinct values for each column specified. If na_rm is TRUE, missing values are removed from the returned vectors of distinct values. Non-existing rows and columns are silently ignored.

Usage
DataBackendDplyr$distinct(rows, cols, na_rm = TRUE)
Arguments
rows

integer()
Row indices.

cols

character()
Column names.

na_rm

logical(1)
Whether to remove NAs or not.

Returns

Named list() of distinct values.

Method missings()

Returns the number of missing values per column in the specified slice of data. Non-existing rows and columns are silently ignored.

Usage
DataBackendDplyr$missings(rows, cols)
Arguments
rows

integer()
Row indices.

cols

character()
Column names.

Returns

Total of missing values per column (named numeric()).

Examples

if (mlr3misc::require_namespaces(c("tibble", "RSQLite", "dbplyr"), quietly = TRUE)) {
  # Backend using a in-memory tibble
  data = tibble::as_tibble(iris)
  data$Sepal.Length[1:30] = NA
  data$row_id = 1:150
  b = DataBackendDplyr$new(data, primary_key = "row_id")

  # Object supports all accessors of DataBackend
  print(b)
  b$nrow
  b$ncol
  b$colnames
  b$data(rows = 100:101, cols = "Species")
  b$distinct(b$rownames, "Species")

  # Classification task using this backend
  task = mlr3::TaskClassif$new(id = "iris_tibble", backend = b, target = "Species")
  print(task)
  head(task)

  # Create a temporary SQLite database
  con = DBI::dbConnect(RSQLite::SQLite(), ":memory:")
  dplyr::copy_to(con, data)
  tbl = dplyr::tbl(con, "data")

  # Define a backend on a subset of the database: do not use column "Sepal.Width"
  tbl = dplyr::select_at(tbl, setdiff(colnames(tbl), "Sepal.Width"))
  tbl = dplyr::filter(tbl, row_id %in% 1:120) # Use only first 120 rows
  b = DataBackendDplyr$new(tbl, primary_key = "row_id")
  print(b)

  # Query disinct values
  b$distinct(b$rownames, "Species")

  # Query number of missing values
  b$missings(b$rownames, b$colnames)

  # Note that SQLite does not support factors, column Species has been converted to character
  lapply(b$head(), class)

  # Cleanup
  rm(tbl)
  DBI::dbDisconnect(con)
}

DataBackend for DuckDB

Description

A mlr3::DataBackend for duckdb. Can be easily constructed with as_duckdb_backend().

Super class

mlr3::DataBackend -> DataBackendDuckDB

Public fields

levels

(named list())
List (named with column names) of factor levels as character(). Used to auto-convert character columns to factor variables.

connector

(⁠function()⁠)
Function which is called to re-connect in case the connection became invalid.

table

(character(1))
Data base table or view to operate on.

Active bindings

table_info

(data.frame())
Data frame as returned by pragma table_info().

rownames

(integer())
Returns vector of all distinct row identifiers, i.e. the contents of the primary key column.

colnames

(character())
Returns vector of all column names, including the primary key column.

nrow

(integer(1))
Number of rows (observations).

ncol

(integer(1))
Number of columns (variables), including the primary key column.

valid

(logical(1))
Returns NA if the data does not inherits from "tbl_sql" (i.e., it is not a real SQL data base). Returns the result of DBI::dbIsValid() otherwise.

Methods

Public methods

Inherited methods

Method new()

Creates a backend for a duckdb::duckdb() database.

Usage
DataBackendDuckDB$new(
  data,
  table,
  primary_key,
  strings_as_factors = TRUE,
  connector = NULL
)
Arguments
data

(connection)
A connection created with DBI::dbConnect(). If constructed manually (and not via the helper function as_duckdb_backend(), make sure that there exists an (unique) index for the key column.

table

(character(1))
Table or view to operate on.

primary_key

(character(1))
Name of the primary key column.

strings_as_factors

(logical(1) || character())
Either a character vector of column names to convert to factors, or a single logical flag: if FALSE, no column will be converted, if TRUE all string columns (except the primary key). For conversion, the backend is queried for distinct values of the respective columns on construction and their levels are stored in ⁠$levels⁠.

connector

(function()⁠)\cr If not ⁠NULL', a function which re-connects to the database in case the connection has become invalid. Database connections can become invalid due to timeouts or if the backend is serialized to the file system and then de-serialized again. This round trip is often performed for parallelization, e.g. to send the objects to remote workers. DBI::dbIsValid() is called to validate the connection. The function must return just the connection, not a dplyr::tbl() object! Note that this this function is serialized together with the backend, including possible sensitive information such as login credentials. These can be retrieved from the stored mlr3::DataBackend/mlr3::Task. To protect your credentials, it is recommended to use the secret package.

Method data()

Returns a slice of the data.

The rows must be addressed as vector of primary key values, columns must be referred to via column names. Queries for rows with no matching row id and queries for columns with no matching column name are silently ignored. Rows are guaranteed to be returned in the same order as rows, columns may be returned in an arbitrary order. Duplicated row ids result in duplicated rows, duplicated column names lead to an exception.

Usage
DataBackendDuckDB$data(rows, cols)
Arguments
rows

integer()
Row indices.

cols

character()
Column names.

Method head()

Retrieve the first n rows.

Usage
DataBackendDuckDB$head(n = 6L)
Arguments
n

(integer(1))
Number of rows.

Returns

data.table::data.table() of the first n rows.

Method distinct()

Returns a named list of vectors of distinct values for each column specified. If na_rm is TRUE, missing values are removed from the returned vectors of distinct values. Non-existing rows and columns are silently ignored.

Usage
DataBackendDuckDB$distinct(rows, cols, na_rm = TRUE)
Arguments
rows

integer()
Row indices.

cols

character()
Column names.

na_rm

logical(1)
Whether to remove NAs or not.

Returns

Named list() of distinct values.

Method missings()

Returns the number of missing values per column in the specified slice of data. Non-existing rows and columns are silently ignored.

Usage
DataBackendDuckDB$missings(rows, cols)
Arguments
rows

integer()
Row indices.

cols

character()
Column names.

Returns

Total of missing values per column (named numeric()).

See Also

https://duckdb.org/

DataBackend for Polars

Description

A mlr3::DataBackend using RPolarsLazyFrame from package polars. Can be easily constructed with as_polars_backend(). mlr3::Tasks can interface out-of-memory files if the polars::RPolarsLazyFrame was imported using a polars::scan_x function. Streaming, a polars alpha feature, is always enabled, but only used when applicable. A connector is not required but can be useful e.g. for scanning larger than memory files

Super class

mlr3::DataBackend -> DataBackendPolars

Public fields

levels

(named list())
List (named with column names) of factor levels as character(). Used to auto-convert character columns to factor variables.

connector

(⁠function()⁠)
Function which is called to re-connect in case the connection became invalid.

Active bindings

rownames

(integer())
Returns vector of all distinct row identifiers, i.e. the contents of the primary key column.

colnames

(character())
Returns vector of all column names, including the primary key column.

nrow

(integer(1))
Number of rows (observations).

ncol

(integer(1))
Number of columns (variables), including the primary key column.

Methods

Public methods

Inherited methods

Method new()

Creates a backend for a polars::RPolarsDataFrame object.

Usage
DataBackendPolars$new(
  data,
  primary_key,
  strings_as_factors = TRUE,
  connector = NULL
)
Arguments
data

(polars::RPolarsLazyFrame)
The data object.

Instead of calling the constructor itself, please call mlr3::as_data_backend() on a polars::RPolarsLazyFrame or polars::RPolarsDataFrame. Note that only polars::RPolarsLazyFrames will be converted to a DataBackendPolars. polars::RPolarsDataFrame objects without lazy execution will be converted to a mlr3::DataBackendDataTable.

primary_key

(character(1))
Name of the primary key column. Because polars does not natively support primary keys, uniqueness of the primary key column is expected but not enforced.

strings_as_factors

(logical(1) || character())
Either a character vector of column names to convert to factors, or a single logical flag: if FALSE, no column will be converted, if TRUE all string columns (except the primary key). For conversion, the backend is queried for distinct values of the respective columns on construction and their levels are stored in ⁠$levels⁠.

connector

(⁠function()⁠)
Optional function which is called to re-connect to e.g. a source file in case the connection became invalid.

Method data()

Returns a slice of the data.

The rows must be addressed as vector of primary key values, columns must be referred to via column names. Queries for rows with no matching row id and queries for columns with no matching column name are silently ignored.

Usage
DataBackendPolars$data(rows, cols)
Arguments
rows

(integer())
Row indices.

cols

(character())
Column names.

Method head()

Retrieve the first n rows.

Usage
DataBackendPolars$head(n = 6L)
Arguments
n

(integer(1))
Number of rows.

Returns

data.table::data.table() of the first n rows.

Method distinct()

Returns a named list of vectors of distinct values for each column specified. If na_rm is TRUE, missing values are removed from the returned vectors of distinct values. Non-existing rows and columns are silently ignored.

Usage
DataBackendPolars$distinct(rows, cols, na_rm = TRUE)
Arguments
rows

(integer())
Row indices.

cols

(character())
Column names.

na_rm

(logical(1))
Whether to remove NAs or not.

Returns

Named list() of distinct values.

Method missings()

Returns the number of missing values per column in the specified slice of data. Non-existing rows and columns are silently ignored.

Usage
DataBackendPolars$missings(rows, cols)
Arguments
rows

(integer())
Row indices.

cols

(character())
Column names.

Returns

Total of missing values per column (named numeric()).

See Also

https://pola-rs.github.io/r-polars/

Examples

if (mlr3misc::require_namespaces("polars", quietly = TRUE)) {
  # Backend using a in-memory data set
  data = iris
  data$Sepal.Length[1:30] = NA
  data$row_id = 1:150
  data = polars::as_polars_lf(data)
  b = DataBackendPolars$new(data, primary_key = "row_id")

  # Object supports all accessors of DataBackend
  print(b)
  b$nrow
  b$ncol
  b$colnames
  b$data(rows = 100:101, cols = "Species")
  b$distinct(b$rownames, "Species")

  # Classification task using this backend
  task = mlr3::TaskClassif$new(id = "iris_polars", backend = b, target = "Species")
  print(task)
  head(task)

  # Write a parquet file to scan
  data$collect()$write_parquet("iris.parquet")
  data = polars::pl$scan_parquet("iris.parquet")

  # Backend that re-reads the parquet file if the connection fails
  b = DataBackendPolars$new(data, "row_id",
    connector = function() polars::pl$scan_parquet("iris.parquet"))
  print(b)

  # Define a backend on a subset of the database: do not use column "Sepal.Width"
  data = data$select(
    polars::pl$col(setdiff(colnames(data), "Sepal.Width"))
  )$filter(
    polars::pl$col("row_id")$is_in(1:120) # Use only first 120 rows
  )

  # Backend with only scanned data
  b = DataBackendPolars$new(data, "row_id", strings_as_factors = TRUE)
  print(b)

  # Query disinct values
  b$distinct(b$rownames, "Species")

  # Query number of missing values
  b$missings(b$rownames, b$colnames)

  # Cleanup
  if (file.exists("iris.parquet")) {
    file.remove("iris.parquet")
  }
}

Convert to DuckDB Backend

Description

Converts to a DataBackendDuckDB using the duckdb database, depending on the input type:

The created backend automatically reconnects to the database if the connection was lost, e.g. because the object was serialized to the filesystem and restored in a different R session. The only requirement is that the path does not change and that the path is accessible on all workers.

Usage

as_duckdb_backend(data, path = getOption("mlr3db.duckdb_dir", ":temp:"), ...)

Arguments

data

(data.frame() | mlr3::DataBackend)
See description.

path

(character(1))
Path for the DuckDB databases. Either a valid path to a directory which will be created if it not exists, or one of the special strings:

  • ":temp:" (default): Temporary directory of the R session is used, see tempdir(). Note that this directory will be removed during the shutdown of the R session. Also note that this usually does not work for parallelization on remote workers. Set to a custom path instead or use special string ":user:" instead.

  • ":user:": User cache directory as returned by R_user_dir() is used.

The default for this argument can be configured via option "mlr3db.sqlite_dir" or "mlr3db.duckdb_dir", respectively. The database files will use the hash of the mlr3::DataBackend as filename with file extension ".duckdb" or ".sqlite". If the database already exists on the file system, the converters will just established a new read-only connection.

...

(any)
Additional arguments, passed to DataBackendDuckDB.

Value

DataBackendDuckDB or mlr3::Task.

Convert to Polars Backend

Description

Converts to a DataBackendPolars using the polars database, depending on the input type:

There is no automatic connection to the origin file set. If the data is obtained using scanning and the data is streamed, a connector can be set manually but is not required.

Usage

as_polars_backend(data, streaming = FALSE, ...)

Arguments

data

(data.frame() | mlr3::DataBackend)
See description.

streaming

(logical(1))
Whether the data should be only scanned (recommended for large data sets) and streamed with every DataBackendPolars operation or loaded into memory completely.

...

(any)
Additional arguments, passed to DataBackendPolars.

Value

DataBackendPolars or mlr3::Task.

Convert to SQLite Backend

Description

Converts to a DataBackendDplyr using a RSQLite database, depending on the input type:

The created backend automatically reconnects to the database if the connection was lost, e.g. because the object was serialized to the filesystem and restored in a different R session. The only requirement is that the path does not change and that the path is accessible on all workers.

Usage

as_sqlite_backend(data, path = getOption("mlr3db.sqlite_dir", ":temp:"), ...)

Arguments

data

(data.frame() | mlr3::DataBackend
See description.

path

(character(1))
Path for the DuckDB databases. Either a valid path to a directory which will be created if it not exists, or one of the special strings:

  • ":temp:" (default): Temporary directory of the R session is used, see tempdir(). Note that this directory will be removed during the shutdown of the R session. Also note that this usually does not work for parallelization on remote workers. Set to a custom path instead or use special string ":user:" instead.

  • ":user:": User cache directory as returned by R_user_dir() is used.

The default for this argument can be configured via option "mlr3db.sqlite_dir" or "mlr3db.duckdb_dir", respectively. The database files will use the hash of the mlr3::DataBackend as filename with file extension ".duckdb" or ".sqlite". If the database already exists on the file system, the converters will just established a new read-only connection.

...

(any)
Additional arguments, passed to DataBackendDplyr.

Value

DataBackendDplyr or mlr3::Task.