import.data: Prepare training and test data.

In sbbrandt/xgbAnalysis: Reaction coordinates analysis

Description

Reads the files coords, states and labels as coordinates, states and labels. The labels represent the column names of the coordinates and the states should match the rows of the coordinates. The data then is separated in a training and a test set and saved in the xgboost matrix format. The following files are created:

• train.xgb.Dmatrix, where the training set is saved.

• test.xgb.Dmatrix, where the test set is saved.

• train.index, where the train indix are saved.

• test.index, where the test index is saved.

• data.import.parameter, where the input parameter are saved to be used by other functions.

Usage

import.data(output_dir, coords, states, labels, trainsplit)

Arguments

output_dir	Character, name of the output directory.
coords	Character, name of the coordinates file
states	Character, name of the states file
labels	Character, filename or 'dihedrals' to get Phi2, Psi2, Phi3, ...
trainsplit	Numeric, fraction of the data that should be used for training, in [0,1]

Examples

import.data("./savefolder","./file.dih", "./macrostates", "./file.dih.names", 0.75)
import.data("./savefolder", "./file.dih", "./macrostates", "dihedrals", 0.5)

sbbrandt/xgbAnalysis documentation built on May 29, 2019, 9:11 a.m.