rjungle: rjungle
In jkruppa/Rjungle: Random Jungle in R

Description Usage Arguments Details Value Author(s) References See Also

Calls the rjungle C++ routines and grows a Random Jungle.

  rjungle(depVarName = "", data = NULL,
    dataFileName = NULL, ntree = 500, mtry = NULL,
    treeType = 1, importance = 1, replace = FALSE,
    proximity = FALSE, keepJungle = TRUE, nthread = 0,
    seed = 123, fileNameOut = character(),
    fileNameIn = character(), balanceData = FALSE,
    verbose = FALSE, convertdata = FALSE,
    inDir = character(), outDir = character(),
    options = "", ...)

`depVarName`	A `character` string providing the name of the dependend variable, if any. If no variable is specified `rjungle` is run unsupervised.
`data`	A `data.frame` containing the data to be analysed.
`dataFileName`	A `character` string defining the name of (or the path to) the dataset to be used for analysis. Not implemented or needed at the moment. See `fileNameIn`.
`ntree`	The `numeric` number of trees to be grown in jungle. Defaults to `500`.
`mtry`	A `numeric`, gives the number of randomly chosen variable sets.
`treeType`	A `numeric` taking values `1`–`5`. Defines what type of regression or classification should be performed. `1`: y(response): nominal, x(input): numeric `2`: y nominal, x nominal `3`: y numeric, x numeric `4`: y numeric, x nominal `5`: as `1`, but recommended for more different values in the input variables (i.e many floating point numbers). Defaults to `1`.
`importance`	A `numeric` taking values `1`–`5`. If `1`, only the GINI-index importance measure is evaluated. If `2`–`5` four other measures will be computed and sorted by the measure coerced with the specified value as follows: `1`: GINI-Index `2`: Breiman-Score `3`: Liaw-Score `4`: raw values, no normalization `5`: Meng-Score
`replace`	A `logical`. If `TRUE`, the set of variables to choose the split from is drawn with replacement, otherwise without replacement. Defaults to `FALSE`.
`proximity`	A `logical`, indicates whether proximites should be computed. Defaults to `FALSE`.
`keepJungle`	A `logical` indicating whether to keep the jungle (e.g. for future use in the call of `predict`). Defaults to `TRUE`.
`nthread`	A `numeric` taking values from `1` to total number of CPUs. Defaults to `0` which is equivalent to `1`.
`seed`	A `numeric` specifing the seed to be used for computation.Defaults to `123`.
`fileNameOut`	A `character` specifing the prefix to the files produced by rjungle. If `missing` the files will be written to `tempfile('rjungledata')`. If specified the files will be written to `getwd` with fileNameOut as prefix to the files. If you want to write to the files to another directory specify the `outDir` parameter.
`fileNameIn`	A `character`, gives the name of the dataset to use if `data` is not specified. Must be a `.dat` file an stored in `getwd` if `inDir` is not given.
`balanceData`	to add
`verbose`	A `logical` indicating whether a progress file shall be created. If `FALSE` no file is created. Defaults to `FALSE`.
`convertdata`	to add
`inDir`	A `character` giving the directory of the input data. Only if `fileNameIn`. Defaults to `getwd`.
`outDir`	A `character` giving the directory where rjungle output data should be stored. Only applies if `fileNameOut` was specified. Defaults to `getwd`.
`options`	Further arguments to be passed to the `rjungle` call.
`...`	Further arguments to be passed to the `system` call.

to add

Returns an object of class rjungle. Contains the setting used in the call of rjungle function and the path to the temporary files produced by rjungle.

Daniel F. Schwarz with modifications by Andreas Bender and Jochen Kruppa

www.randomjungle.de

importance, confusion

jkruppa/Rjungle documentation built on May 19, 2019, 12:45 p.m.