split_ts: Automatic Time Series Cross-Validation split

In opoyc/autoforecast: Forecast Development Kit

Description

Time series CV split heuristics should keep temporal dependencies, henceforth, the sample should not be "shuffled" into train and test. This function uses a different strategy to define the test and train sets maintain the order of the data.

Usage

split_ts(.data, test_size, lag)

Arguments

.data	DataFrame, tibble or tsibble structures.
test_size	Numeric. How many periods will be use to asses the forecast accuracy.
lag	Numeric. How many periods ahead to start the test size.

Value

Nested data-frames or tibbles.

Examples

## Not run: 
split_ts()

## End(Not run)

opoyc/autoforecast documentation built on May 18, 2021, 1:29 a.m.