Reading data from different sources in their original format.

Description

The package contains functions, which allow to maintain and use a structure describing a collection of machine learning datasets and read them into R environment using a unified interface, see function prepareDSList() and dsRead().

Details

The data are not part of the package. The package requires to receive a path to a local copy of the data and their description. The description of the data sets consists of a directory, which contains an XML file contents.xml and subdirectory "scripts", which contains an R script for each data set, which reads the data set into R. File contents.xml contains information on all the data sets. In particular it contains their names for local identification, their public names, and the names of files representing the data set. The name of the script for reading a data set is derived from its identification name. The complete list of the fields in contents.xml may be obtained using getFields().

For the simplest use of the package for reading the data sets, the functions prepareDSList() and dsRead() are sufficient. The remaining functions are useful for including further data sets to the description. Use help(package=readMLData) or library(help=readMLData) to see the list of functions.

The list of fields, which should be included in "contents.xml", consists of the fields with either usage=="obligatory" or usage=="optional" in the table produced by getFields(). Fields with usage=="additional" and usage=="computed" are included automatically by the function prepareDSList().

An example of the description directory describing three UCI data sets is in exampleDescription subdirectory of the installed package. The data themselves are in exampleData subdirectory. See http://www.cs.cas.cz/~savicky/readMLData/ for description files of further data sets from UCI Machine Learning Repository.

Author(s)

Petr Savicky

References

UCI Machine Learning Repository, http://archive.ics.uci.edu/ml/.

Additional resources for the CRAN package readMLData, http://www.cs.cas.cz/~savicky/readMLData/.

Determine the type of values in each column of a data frame.

Description

For each column, its class and the number of different values is determined. For numeric columns, also the minimum and maximum is computed.

Usage

analyzeData(dat)

Arguments

Value

A data frame with columns "class", "num.unique", "min", "max", which correspond to properties of columns of dat. The rows in the output data frame correspond to the columns of dat.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  dat <- dsRead(dsList, "glass")
  analyzeData(dat)

Checks consistency of the data frame dsList.

Description

Checks consistency of the parameters specified for each dataset in the dsList data frame created by prepareDSList().

Usage

checkConsistency(dsList, outputInd=FALSE)

Arguments

Value

Depending on outputInd, either a vector of indices of data sets with a conflict between the specified parameters or NULL invisibly.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  checkConsistency(dsList)

Compares the type of columns stored in dsList and in a data set itself.

Description

Compares types.

Usage

checkType(dsList, id, dat=NULL)

Arguments

Value

The name of the tested data set and the result of the test is printed. If errors are found, a more detailed message is printed. The output value is TRUE or FALSE invisibly according, whether the types are correct or not.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  checkType(dsList, 1)

Run an external tool to download a data set.

Description

The function allows to run an external download tool with arguments read from a file in a data folder.

Usage

dsDownload(dsList, id, command, fileName)

Arguments

Details

If no data set or more than one data set corresponding to id is found, a corresponding error message is printed.

Value

Function has no value. The protocol generated by the specified tool is printed.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

## Not run: 
  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  dat <- dsDownload(dsList, "glass", "wget", "links.txt")

## End(Not run)

Loading machine learning data from a directory tree using a unified interface.

Description

The function allows to read data sets included in the description in the data frame dsList into R environment using a unified interface.

Usage

dsRead(dsList, id, responseName = NULL, originalNames=TRUE,
deleteUnused=TRUE, keepContents=FALSE)

Arguments

Details

The function uses dsList$avaiable to determine, whether the files for the required data set is present in the local directory dsList$pathData. If not, a corresponding error message is printed. See prepareDSList() and getAvailable().

Value

A data frame containing the required data set, possibly transformed according to the setting of the parameters responseName, originalNames, deleteUnused. If an error occurred, the function outputs NULL.

Author(s)

Petr Savicky

See Also

readMLData, prepareDSList, getAvailable.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  dat <- dsRead(dsList, "glass")
  dim(dat)

Search a dataset by string matching against the names stored in dsList.

Description

The function allows string matching against some of the fields "identification", "fullName", "dirName", "files" of the structure describing the data sets.

Usage

dsSearch(dsList, id, searchField=c("identification", "fullName", "dirName", "files"),
            searchType=c("exact", "prefix", "suffix", "anywhere"), caseSensitive=FALSE)

Arguments

Details

The parameter searchField determines, which column of dsList is searched, parameters searchType and caseSensitive influence the type of search. These three parameters are ignored, if id is numeric.

Regular expressions are not used. Matching with searchType="exact" is done with ==, searchType="prefix" and searchType="suffix" are implemented using substr(), searchType="anywhere" is implemented using grep(, fixed=TRUE).

Value

Data frame containing the indices and identification of the matching data sets and the value of the search field, if applicable.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  dsSearch(dsList, "ident", searchField="fullName", searchType="anywhere")

Sort the rows of a data frame.

Description

Sort the rows of a data frame lexicographically. This allows to compare two data sets as sets of cases disregarding their order.

Usage

dsSort(dat)

Arguments

Details

The function calls order() with the columns of dat as the sorting criteria.

Value

Data frame, whose rows are reordered by the sorting.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  dat <- dsRead(dsList, "glass")
  sorted <- dsSort(dat)

Checks consistency of the data frame dsList.

Description

Checks whether all the files of a specified data set are accesible in a local directory.

Usage

getAvailable(dsList, id=NULL, asLogical=FALSE)

Arguments

Details

The test is not completely reliable, since it only verifies that the files with the required file name are accessible. If the files require some transformations after download and these are not performed, the data set is still reported as available. The test uses file names specified in contents.xml file. If these names are by mistake different from the files actually read in the reading scripts, then the test may also yield an incorrect result.

Value

Logical vector of the length length(id) specifying for each component of id the result of the check or a character vector containing the identification of the available data sets.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  getAvailable(dsList)

Prints the information on the fields in the data frame dsList describing the data sets.

Description

The data frame dsList contains names of the data sets, the names of the directories, the files, which belong to each of the data sets, and some other information. The function returns a table describing the fields and their usage.

Usage

getFields()

Value

Table containing the names, types and usage of the fields expected in dsList.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  getFields()

Determine the path to package example directories.

Description

Appends the path to the directory of an installed package and a name of its subdirectory.

Usage

getPath(dirName)

Arguments

Value

Character string, which is a full path to the required example directory in an installed package.

Author(s)

Petr Savicky

See Also

prepareDSList

Determines the type vector for an input data set.

Description

The type information is derived from the contents of individual columns of an input data frame.

Usage

getType(dat)

Arguments

Value

A character vector of length ncol(dat) containing "n" for numerical columns, the number of different values for character or factor columns, and "o" otherwise.

Author(s)

Petr Savicky

See Also

readMLData.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)
  dat <- dsRead(dsList, "annealing")
  getType(dat)

Prepares a data frame dsList, which describes the data contained in a local data description directory.

Description

The data frame dsList is needed to read the data contained in a directory tree below dsList$pathData using dsRead(). The directory pathDescription is expected to contain the file contents.xml and subdirectory scripts with R scripts for reading the data sets.

Usage

prepareDSList(pathData, pathDescription)

Arguments

Details

The character "~" expands to your home directory.

The directory pathData need not contain all the data sets included in pathDescription/contents.xml. The function getAvailable() is called and its output is stored in column availability of the output data frame, which is logical and specifies for each data set, whether it is or is not present.

See http://www.cs.cas.cz/~savicky/readMLData/ for description files of some of the data sets from UCI Machine Learning Repository. See the help page readMLData for more information on the structure of the description files.

Value

Data frame with columns pathData, pathDescription, and other as listed by getFields(). The output data frame can be used as dsList parametr of functions dsSearch(), dsRead(), checkConsistency(), checkType().

Author(s)

Petr Savicky

See Also

readMLData, getAvailable, checkConsistency.

Examples

  pathData <- getPath("exampleData")
  pathDescription <- getPath("exampleDescription")
  dsList <- prepareDSList(pathData, pathDescription)

Handling XML files.

Description

Input and output of a data set description from and to a XML file. These functions are not inteded for direct use by the user for reading the data sets. The function readDSListFromXML() is called from prepareDataDir(). The function saveDSListAsXML is used for preparing the file contents.xml in the data set description directory.

Usage

readDSListFromXML(filename)
saveDSListAsXML(dsList, filename)

Arguments

Value

saveDSListAsXML() returns the filename of the created file. readDSListFromXML() returns a data frame with the description of the data sets.

Author(s)

Petr Savicky

See Also

readMLData.