Package 'segen'

Title: Sequence Generalization Through Similarity Network
Description: Proposes an application for sequence prediction generalizing the similarity within the network of previous sequences.
Authors: Giancarlo Vercellino [aut, cre, cph]
Maintainer: Giancarlo Vercellino <[email protected]>
License: GPL-3
Version: 2.0.0
Built: 2026-05-19 08:24:57 UTC
Source: https://github.com/cran/segen

Help Index

segen

Description

Sequence Generalization Through Similarity Network

Usage

segen(
  df,
  seq_len = NULL,
  similarity = NULL,
  dist_method = NULL,
  rescale = NULL,
  smoother = FALSE,
  ci = 0.8,
  error_scale = "naive",
  error_benchmark = "naive",
  n_windows = 10,
  n_samp = 30,
  dates = NULL,
  seed = 42,
  use_parallel = FALSE,
  parallel_workers = NULL
)

Arguments

df

data.frame of time features (all numeric OR all categorical).
seq_len

integer, forecasting horizon. If NULL, auto-sampled.
similarity

numeric in (0,1), similarity quantile. If NULL, sampled.
dist_method

character. Options: "euclidean","manhattan","maximum","minkowski","correlation","dtw". If NULL, sampled from available methods (skips 'dtw' if pkg missing).
rescale

logical, rescale weights before normalization.
smoother

logical, apply loess smoothing for numeric features.
ci

numeric in (0,1), confidence level.
error_scale

"naive" or "deviation".
error_benchmark

"naive" or "average".
n_windows

integer, rolling validation windows.
n_samp

integer, random search samples.
dates

Date vector aligned with rows of df (optional).
seed

integer, RNG seed.
use_parallel

logical, use furrr/future for parallel exploration.
parallel_workers

NULL or integer, number of workers when parallel.

Value

list with exploration, history, best_model, time_log.

This function returns a list including:

  • exploration: list of all not-null models, complete with predictions and error metrics

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

  • best_model: results for the best selected model according to the weighted average rank, including:

    • predictions: for continuous variables, min, max, q25, q50, q75, quantiles at selected ci, mean, sd, mode, skewness, kurtosis, IQR to range, risk ratio, upside probability and divergence for each point fo predicted sequences; for factor variables, min, max, q25, q50, q75, quantiles at selected ci, proportions, difformity (deviation of proportions normalized over the maximum possible deviation), entropy, upgrade probability and divergence for each point fo predicted sequences

    • testing_errors: testing errors for each time feature for the best selected model (for continuous variables: me, mae, mse, rmsse, mpe, mape, rmae, rrmse, rame, mase, smse, sce, gmrae; for factor variables: czekanowski, tanimoto, cosine, hassebrook, jaccard, dice, canberra, gower, lorentzian, clark)

    • plots: standard plots with confidence interval for each time feature

  • time_log

Author(s)

Giancarlo Vercellino [email protected]

Maintainer: Giancarlo Vercellino [email protected] [copyright holder]

See Also

Useful links:

Examples

segen(time_features[, 1, drop = FALSE], seq_len = 30, similarity = 0.7, n_windows = 3, n_samp = 1)

time features example: IBM and Microsoft Close Prices

Description

A data frame with with daily with daily prices for IBM and Microsoft since April 2020

Usage

time_features

Format

A data frame with 2 columns and 1324 rows.

Source

finance.yahoo.com