sliding_ase_univariate.Rd
Function to calculate the sliding window ASE for a model Supports ARMA, ARIMA, ARUMA (seasonal ARIMA) and Signal Plus Noise Models
sliding_ase_univariate( x, phi = 0, theta = 0, d = 0, s = 0, linear = NA, freq = NA, n.ahead = NA, batch_size = NA, step_n.ahead = TRUE, verbose = 0, ... )
x | time series realization |
---|---|
phi | phi values associated with the ARIMA model |
theta | theta values associated with the ARIMA model |
d | differencing 'd' associated with the ARIMA model |
s | seasonality 's' of the ARIMA model |
linear | (TRUE|FALSE) If using a Signal Plus Noise model, should a linear signal be used? |
freq | If using a sinusoidal signal, what is the frequency of the signal? |
n.ahead | last n.ahead data points in each batch will be used for prediction and ASE calculations |
batch_size | Window Size used |
step_n.ahead | Whether to step each batch by n.ahead values (Default = FALSE) |
verbose | How much to print during the model building and other processes (Default = 0) |
... | any additional arguments to be passed to the forecast functions (e.g. max.p for sigplusnoise model, lambda for ARUMA models) |
Named list 'ASEs' - ASE values 'time_test_start' - Time Index indicating start of test time corresponding to the ASE values 'time_test_end' - Time Index indicating end of test time corresponding to the ASE values 'batch_num' - Indicates the batch number for each ASE value 'f' - Forecasts for each batch 'll' - Lower Forecast Limit for each batch 'ul' - Upper Forecast Limit for each batch 'time.forecasts' - Time Corresponding to each forecast, upper and lower limit values