data_prep_func: Prepares data for modeling
In vidarsumo/sumots: Helper functions for Sumo Analytics

Prepares data for modeling

data_prep_func(
  data,
  outcome_var,
  negative_to_zero = FALSE,
  fix_gap_size = FALSE,
  max_gap_size,
  remove_one_obs = FALSE,
  trailing_zero = FALSE,
  transformation = "none",
  use_holidays = FALSE,
  holidays_to_use_1 = NULL,
  holidays_to_use_2 = NULL,
  use_seasonal_lag = TRUE,
  seasonal_frequency,
  horizon,
  clean = FALSE,
  drop_na_values = TRUE,
  use_holiday_to_clean = FALSE,
  holiday_for_clean = NULL,
  use_abc_category = FALSE,
  pacf_threshold = 0.2,
  no_fourier_terms = 5,
  fourier_k = 5,
  slidify_period = c(4, 8),
  use_own_fourier = FALSE,
  fourier_terms,
  intermittent = 0.3,
  recursive_data = FALSE,
  no_recursive_lag,
  xreg = NULL,
  fill_na_with_zero = TRUE,
  anomaly = FALSE
)

`data`	Data frame with data on the right date format e.g. daily, weekly, monthly and with column named 'id'
`outcome_var`	Name of the outcome variable. The function will change the outcome variable name to 'outcome'
`negative_to_zero`	Recodes negative values as zero, defaults to TRUE
`max_gap_size`	The maximum length that the outcome can be zero. If the interval is larger than max_gap_size then only use data after the interval
`trailing_zero`	Extends all time series, back and forward, so they will be the same length. Defaults to FALSE
`transformation`	Should the series be transformed, e.g. log or log1p. Defaults to none
`use_holidays`	Should national holidays be included. As of now this has to be a dataframe supplied by the user to the holidays_to_use argument
`holidays_to_use_1`	Data frame of dummy holidays. Outcome of the create_holiday() function: fridagar_tbl
`holidays_to_use_2`	Data frame of holidays, one variable.
`use_seasonal_lag`	Should lag of outcome variable, equal to the seasonality, be used. Defaults to TRUE.
`seasonal_frequency`	The frequency of the data. E.g. 52 for weekly data
`horizon`	The forecast horizon
`clean`	Should the data be cleand for outliers. Defaults to FALSE
`drop_na_values`	When creating data_prepared_tbl, should NA's be dropped. Defaults to TRUE
`use_holiday_to_clean`	Uses fridagar_one_var from the create_holiday() function to revert series to original value if cleand
`pacf_threshold`	Threshold for where to cut the PACF to choose terms for the fourier calculation
`no_fourier_terms`	Number of fourier terms, defultas to 5
`fourier_k`	The fourier term order, defaults to 5
`slidify_period`	The window size, defaults to c(4, 8)
`use_own_fourier`	Should you use your own fourier terms? Defaults to FALSE
`fourier_terms`	The fourier terms to include.
`intermittent`	Intermittent threshold when to remove anomaly label. Defaults to 0.3
`recursive_data`	Should the data be prepared for a recursive forecasting. Defaults to FALSE.
`no_recursive_lag`	The number of lags to be.
`xreg`	External regressors to add
`fill_na_with_zero`	Used when drop_na = FALSE to fill missing values with zero instead of dropping them.
`anomaly`	Should anomaly detection variable added. Defaults to FALSE.