tetragon {tetragon}

tetragon

Description

Each sequence is predicted by expanding the distance matrix. The compact set of hyper-parameters is tuned via grid or random search.

Usage

tetragon(
  df,
  seq_len = NULL,
  smoother = F,
  ci = 0.8,
  method = NULL,
  distr = NULL,
  n_windows = 3,
  n_sample = 30,
  dates = NULL,
  error_scale = "naive",
  error_benchmark = "naive",
  seed = 42
)

Arguments

df

A data frame with time features as columns. They could be continuous variables or not.

seq_len

Positive integer. Time-step number of the projected sequence. Default: NULL (random selection between maximum boundaries).

smoother

Logical. Perform optimal smoothing using standard loess. Default: FALSE

ci

Confidence interval. Default: 0.8.

method

String. Distance method for calculating distance matrix among sequences. Options are: "euclidean", "manhattan", "maximum", "minkowski". Default: NULL (random selection among all possible options).

distr

String. Distribution used to expand the distance matrix. Options are: "norm", "logis", "t", "exp", "chisq". Default: NULL (random selection among all possible options).

n_windows

Positive integer. Number of validation tests to measure/sample error. Default: 3 (but a larger value is strongly suggested to really understand your accuracy).

n_sample

Positive integer. Number of samples for random search. Default: 30.

dates

Date. Vector with dates for time features.

error_scale

String. Scale for the scaled error metrics (only for continuous variables). Two options: "naive" (average of naive one-step absolute error for the historical series) or "deviation" (standard error of the historical series). Default: "naive".

error_benchmark

String. Benchmark for the relative error metrics (only for continuous variables). Two options: "naive" (sequential extension of last value) or "average" (mean value of true sequence). Default: "naive".

seed

Positive integer. Random seed. Default: 42.

Value

This function returns a list including:

• exploration: list of all explored models, complete with predictions, testing metrics and plots

• history: a table with the sampled models, hyper-parameters, validation errors

• best: results for the best model including:

  • predictions: min, max, q25, q50, q75, quantiles at selected ci, and a bunch of specific measures for each point fo predicted sequences

  • testing_errors: testing errors for one-step and sequence for each ts feature

  • plots: confidence interval plot for each time feature

• time_log

Author(s)

Giancarlo Vercellino [e-mail]

See Also

Useful links:

• https://rpubs.com/giancarlo_vercellino/tetragon

Examples

tetragon(covid_in_europe[, c(2, 4)], seq_len = 40, n_sample = 2)