split_train_test_set: Split the data into a training, validation and test set

In xavierkamp/tsForecastR: Forecasting with Traditional Time Series and Machine Learning Models

Description

Split the data into a training, validation and test set

Usage

split_train_test_set(
  mts_data,
  fc_horizon = 12,
  bt_iter = 1,
  valid_set_size = 0,
  tmp_test_set_size = 0,
  backtesting_opt = NULL,
  ...
)

Arguments

mts_data	A univariate or multivariate 'ts', 'mts' or 'xts' object
fc_horizon	An integer, the forecasting horizon (i.e. the number of periods to forecast)
bt_iter	An integer, number of the current backtesting operation (i.e. forecasting exercise). This argument must be smaller or equal to the number of backtesting operations to perform.
valid_set_size	An integer, the validation set size (default = 0)
tmp_test_set_size	An integer, size of a second test set (used by h2o.automl) (default = 0)
backtesting_opt	A list, options which define the backtesting approach: use_bt - A boolean, to determine whether forecasts should be generated on future dates (default) or on past values. Generating forecasts on past dates allows to measure past forecast accuracy and to monitor a statistical model's ability to learn signals from the data. nb_iters - An integer, to determine the number of forecasting operations to apply (When no backtesting is selected, then only one forecasting exercise is performed) method - A string, to determine whether to apply a 'rolling' (default) or a 'moving' forecasting window. When 'rolling' is selected, after each forecasting exercise, the forecasting interval increments by one period and drops the last period to include it in the new training sample. When 'moving' is selected, the forecasting interval increments by its size rather than one period. sample_size - A string, to determine whether the training set size should be 'expanding' (default) or 'fixed'. When 'expanding' is selected, then after each forecasting operation, the periods dropped from the forecasting interval will be added to the training set. When 'fixed' is selected, then adding new periods to the training set will require dropping as many last periods to keep the set's size constant.
...	Additional arguments to be passed to the function

Value

A list with training, validation and test sets

xavierkamp/tsForecastR documentation built on Feb. 1, 2020, 10:16 a.m.