feature_engineering_ts: Automatic Time Series Feature Engineering

In opoyc/autoforecast: Forecast Development Kit

Description

This function applies different heuristics to add time series features to the original data.

Usage

feature_engineering_ts(
  .data,
  lag_var = list(),
  ma_var = list(),
  numeric_seas = FALSE,
  hierarchy_seas = FALSE
)

Arguments

.data	data-frame or tibble
lag_var	list: defines the number of lag periods for a given variable, for instance, list(y_var = 1) will generate a lag (autoregressive) variable of 1 period.
ma_var	list: defines the number of periods before and after to apply a mean, for instance, list(y_var = c(1, 2)) will generate a moving average of 1 period before and 2 after a given point.
numeric_seas	logical: whether or not to print numberic seasonal variables instead of factor.
hierarchy_seas	logical: whether or not to provide higher aggregation levels. For instance, a time series of weekly frequency will generate besides week_seas, also month_seas, and year_seas.

Value

data-frame or tibble

Examples

## Not run: 
feature_engineering_ts()

## End(Not run)

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