make.ml.sets: Make training/dev/test test

In pierrecattin/thesis-resources: Thesis Resources

Description

Removes incomplete observations and split the time series into distinct sets.

Usage

make.ml.sets(x, y, training.start, training.end, dev.end, test.end,
  train.for.test.start, scale = T)

Arguments

x	: xts of features
y	: xts of true classes
training.start	: date
training.end	: date
dev.end	: date
test.end	: date. Optional, if missing, no test set is created
test.train.start	: date. Optional. Start of the period used to train the model for the test set
scale:	inidcates if x should be scaled to mean 0 and sd 1 (based on training sample mean and sd)

Value

list containing x.train(xts), x.dev(xts), (x.test(xts)), y.train(factor), y.dev(factor), (y.test(factor))

pierrecattin/thesis-resources documentation built on May 4, 2019, 8:42 a.m.