TrainAndTest.break: Create a data.table for actual v fitted, with train and test...

In Braja93/braja: General data manipulation functions

Description

Takes a lm object and outputs an actual vs fitted table.

Usage

TrainAndTest.break(data, variable, date.index = "Week", to.break = NA,
  append = FALSE)

Arguments

data	data.table.
variable	Character. The variable name, of which to create training and test sets.
date.index	Character. The date index of the lm(data = ?) variable.
to.break	Defaults to NA which creates the training set as 2/3 of the data, the last 1/3 is the test. Set to a number between 0 and 1 to act as a percentage split, e.g. 0.5 - half the data is training. Or, enter a character in yyyy-mm-dd as the date to split.
append	Defaults to FALSE - which returns date.index, and the training and test variables. TRUE returns data, with two new columns at the end - useful to overwrite the original table.

Examples

x <- data.table::fread(system.file("extdata", "lm.csv", package = "nladwa"))
x[, Week := excelDateToRDate(Week)]

x <- TrainAndTest.break(x, "TotalGrossLeads", to.break = "2015-04-06", append = TRUE)
# 06/04/2015

lm1 <- lm(TotalGrossLeads.Train ~ Search.Spend + Competitor.Spend, data = x)
lm1.trainAndTest <- lm.trainAndTest(lm.object = lm1, date.index = "Week")
print(lm1.trainAndTest)

lm.trainAndTest.ggplot(lm1.trainAndTest)
lm.trainAndTest.resid.ggplot(lm1.trainAndTest)

MAPE.trainAndTest(lm1.trainAndTest)

Braja93/braja documentation built on May 6, 2019, 12:07 a.m.