modify_pred: Modify the prediction by auto-adjusting
In placeboo/amplify: Forecasting Daily Time Seriers with Double Seasonal Patterns

Modify the prediction by auto-adjusting

modify_pred(
  data,
  vars = c("date", "y"),
  time.window = NULL,
  par = list(lambda = 1.5, n = NULL),
  holiday = NULL,
  weekday.sep = FALSE
)

`data`	A data framem includes a column called `date`.
`vars`	A charater vector represents the names of date and the numeric vector of proportion error.
`time.window`	A numeric vector shows the begining date and ending date, see `build_date`.
`par`	A list of nuisance parameters. `lambda=1.5`, a number decribes the "distance" between the outliers and numeric variable `var`. `n`, the threshold number of times the fixed date or the fixed weekday is treated as outliers. If `n = NULL`, it takes the half number of years.
`holiday`	A data frame includes `date` and `name` columns. If users want to adjust more holidays, simply add the date and name of the holidays. The `date` should be in date format, and the `name` should be charater.
`weekday.sep`	A logical value indicating whether modify the value separately by weekdays

A list includes:

h.mat A data frame with two columns date and h.hat. h.hat is the deducating/increasing proportion of estimated time series.
fix.date, fix.weekday, holiday. Proprotion erorr for the past several years.