Title:	Comprehensive Regime Change Detection in Time Series
Version:	0.1.1
Description:	A unified framework for detecting regime changes (changepoints) in time series data. Implements both frequentist methods including Cumulative Sum (CUSUM, Page (1954) <doi:10.1093/biomet/41.1-2.100>), Pruned Exact Linear Time (PELT, Killick, Fearnhead, and Eckley (2012) <doi:10.1080/01621459.2012.737745>), Binary Segmentation, and Wild Binary Segmentation, as well as Bayesian methods such as Bayesian Online Changepoint Detection (BOCPD, Adams and MacKay (2007) <doi:10.48550/arXiv.0710.3742> and Shiryaev-Roberts. Supports offline analysis for retrospective detection and online monitoring for real-time surveillance. Provides rigorous uncertainty quantification through confidence intervals and posterior distributions. Handles univariate and multivariate series with detection of changes in mean, variance, trend, and distributional properties.
License:	MIT + file LICENSE
Encoding:	UTF-8
RoxygenNote:	7.3.3
Depends:	R (≥ 4.0.0)
Imports:	stats, ggplot2 (≥ 3.4.0), rlang (≥ 1.1.0), cli (≥ 3.6.0), magrittr
Suggests:	testthat (≥ 3.0.0), knitr, rmarkdown, plotly, patchwork, covr, JuliaCall (≥ 0.17.0), keras (≥ 2.9.0), tensorflow (≥ 2.9.0), reticulate (≥ 1.26)
VignetteBuilder:	knitr
Config/testthat/edition:	3
URL:	https://github.com/IsadoreNabi/RegimeChange
BugReports:	https://github.com/IsadoreNabi/RegimeChange/issues
LazyData:	true
SystemRequirements:	Julia (>= 1.6) for optional high-performance backend, Python (>= 3.8) with TensorFlow for deep learning methods
NeedsCompilation:	no
Packaged:	2026-02-11 19:07:44 UTC; ROG
Author:	José Mauricio Gómez Julián [aut, cre]
Maintainer:	José Mauricio Gómez Julián <isadorenabi@pm.me>
Repository:	CRAN
Date/Publication:	2026-02-13 16:34:59 UTC

RegimeChange: Comprehensive Regime Change Detection in Time Series

Description

A unified framework for detecting regime changes (changepoints) in time series data. Implements both frequentist methods including Cumulative Sum (CUSUM, Page (1954) doi:10.1093/biomet/41.1-2.100), Pruned Exact Linear Time (PELT, Killick, Fearnhead, and Eckley (2012) doi:10.1080/01621459.2012.737745), Binary Segmentation, and Wild Binary Segmentation, as well as Bayesian methods such as Bayesian Online Changepoint Detection (BOCPD, Adams and MacKay (2007) https://arxiv.org/abs/0710.3742) and Shiryaev-Roberts. Supports offline analysis for retrospective detection and online monitoring for real-time surveillance. Provides rigorous uncertainty quantification through confidence intervals and posterior distributions. Handles univariate and multivariate series with detection of changes in mean, variance, trend, and distributional properties.

Author(s)

Maintainer: José Mauricio Gómez Julián isadorenabi@pm.me (ORCID)

See Also

Useful links:

• https://github.com/IsadoreNabi/RegimeChange

• Report bugs at https://github.com/IsadoreNabi/RegimeChange/issues

Adjusted Rand Index

Description

Rand Index corrected for chance agreement.

Usage

adjusted_rand_index(detected, true_cp, n)

Arguments

Value

Adjusted Rand Index (can be negative, 1 is perfect)

Autoencoder-based Changepoint Detection

Description

Detects changepoints by identifying regions where reconstruction error is anomalously high, indicating the model (trained on normal patterns) fails to reconstruct the data.

Usage

autoencoder_detect(
  data,
  window_size = 50,
  latent_dim = 10,
  hidden_dims = c(32, 16),
  epochs = 100,
  batch_size = 32,
  threshold = NULL,
  contamination = 0.1,
  variational = FALSE,
  verbose = FALSE
)

Arguments

Value

List with:

Examples

## Not run: 
data <- c(rnorm(100), rnorm(100, mean = 3), rnorm(100))
result <- autoencoder_detect(data, window_size = 30)
plot(result$reconstruction_error, type = "l")
abline(v = result$changepoints, col = "red")

## End(Not run)

Benchmark Julia vs R Performance

Description

Compare execution time between Julia and R implementations.

Usage

benchmark_backends(data, method = "pelt", n_reps = 10, ...)

Arguments

Value

Data frame with timing results (invisibly).

Examples

data <- c(rnorm(500), rnorm(500, 3))
benchmark_backends(data, method = "pelt", penalty = log(1000))

Binary Segmentation Changepoint Detection

Description

Recursive binary segmentation for changepoint detection. Greedy algorithm that recursively splits at the best changepoint.

Usage

binary_segmentation(
  data,
  type = "both",
  penalty = "BIC",
  min_segment = 2,
  n_changepoints = "multiple",
  threshold = NULL,
  ...
)

Arguments

Value

List with changepoints

Examples

data <- c(rnorm(100), rnorm(100, mean = 2))
result <- binary_segmentation(data)

Bayesian Online Changepoint Detection

Description

Implements the BOCPD algorithm of Adams and MacKay (2007). Maintains a posterior distribution over the run length (time since last changepoint) and updates it online as new observations arrive.

Usage

bocpd(
  data,
  type = "both",
  prior = NULL,
  hazard = NULL,
  threshold = 0.3,
  truncate_run_length = NULL,
  ...
)

Arguments

Value

List containing:

• changepoints: Detected changepoint locations

• posterior: Matrix of run length posteriors over time

• prob_change: Probability of changepoint at each time

• map_run_length: Maximum a posteriori run length at each time

References

Adams, R. P. and MacKay, D. J. C. (2007). Bayesian Online Changepoint Detection. arXiv:0710.3742

Examples

data <- c(rnorm(100, mean = 0), rnorm(100, mean = 3))

result <- bocpd(data)

result <- bocpd(data, prior = normal_gamma(mu0 = 0, kappa0 = 1,
                                            alpha0 = 1, beta0 = 1))


result <- bocpd(data, hazard = 0.01)

Compare Multiple Detection Methods

Description

Runs multiple detection methods on the same data and compares results.

Usage

compare_methods(
  data,
  methods = c("pelt", "binseg", "wbs", "bocpd"),
  true_changepoints = NULL,
  tolerance = 5,
  ...
)

Arguments

Value

Object of class "regime_comparison"

Examples

true_cp <- c(50, 100)
data <- c(rnorm(50), rnorm(50, mean = 2), rnorm(50))

comparison <- compare_methods(
  data,
  methods = c("pelt", "binseg"),
  true_changepoints = true_cp
)

Constant Hazard Prior (Alias for Geometric)

Description

Constant Hazard Prior (Alias for Geometric)

Usage

constant_hazard(lambda = 0.01)

Arguments

Value

An object of class "hazard_prior"

Covering Metric

Description

Measures how well detected segments cover true segments using IoU.

Usage

covering_metric(detected, true_cp, n)

Arguments

Value

Covering metric (0 to 1)

Contrastive Predictive Coding for Changepoint Detection

Description

Uses self-supervised contrastive learning to detect changepoints by identifying where the learned representations change significantly.

Usage

cpc_detect(
  data,
  window_size = 64,
  encoding_dim = 32,
  n_negative = 10,
  prediction_steps = 5,
  epochs = 100,
  threshold = NULL,
  verbose = FALSE
)

Arguments

Value

List with changepoints and learned encodings

References

Oord, A. v. d., Li, Y., and Vinyals, O. (2018). Representation Learning with Contrastive Predictive Coding

CROPS - Changepoints for a Range of Penalties

Description

Efficiently computes optimal segmentations for a range of penalty values, useful for model selection.

Usage

crops_detect(
  data,
  penalty_range = c(0.1, 100),
  method = "pelt",
  min_segment = 2
)

Arguments

Value

List with:

References

Haynes, K., Eckley, I. A., and Fearnhead, P. (2017). Computationally efficient changepoint detection for a range of penalties. Journal of Computational and Graphical Statistics, 26(1), 134-143.

CUSUM Changepoint Detection

Description

Detects changepoints using the Cumulative Sum (CUSUM) statistic. Foundational method for detecting changes in mean.

Usage

cusum(
  data,
  type = "mean",
  threshold = 4,
  mode = "offline",
  mu0 = NULL,
  sigma = NULL,
  ...
)

Arguments

Value

List with changepoints and statistics

Examples

data <- c(rnorm(50), rnorm(50, mean = 2))
result <- cusum(data)

Detect changepoints using enhanced PELT algorithm

Description

Pruned Exact Linear Time (PELT) algorithm for optimal changepoint detection with enhancements for autocorrelated data and numerical stability.

Usage

detect_pelt(
  data,
  type = "both",
  penalty = "MBIC",
  min_segment = 2,
  robust = FALSE,
  correct_ar = FALSE,
  merge_close = NULL,
  ...
)

Arguments

Details

This implementation includes enhancements over standard PELT:

Pre-whitening for autocorrelated data: When correct_ar = TRUE, the function estimates the AR(1) coefficient and transforms the data via y'_t = y_t - \rho y_{t-1} before applying PELT.

Configurable robustness: The robust parameter accepts:

• FALSE: Standard MLE-based estimation

• TRUE or "moderate": 10\

• "mild": 5\

• "aggressive": 15\

• "auto": Automatically select based on data characteristics

Value

A list with components:

changepoints

Integer vector of detected changepoint locations

n_changepoints

Number of changepoints detected

cost

Final cost value

ar_coefficient

Estimated AR(1) coefficient (if correct_ar = TRUE)

prewhitened

Logical indicating if pre-whitening was applied

robust

Logical indicating if robust estimation was used

robust_level

Character indicating robustness level used

information_criterion

Reserved for future use

References

Killick, R., Fearnhead, P., and Eckley, I. A. (2012). Optimal detection of changepoints with a linear computational cost. JASA, 107(500), 1590-1598.

Examples

data <- c(rnorm(100, 0), rnorm(100, 3))
result <- detect_pelt(data, type = "mean")


data_contaminated <- c(rnorm(100, 0), rnorm(100, 3))
data_contaminated[sample(200, 10)] <- rnorm(10, 0, 10)
result_robust <- detect_pelt(data_contaminated, robust = TRUE)


result_auto <- detect_pelt(data_contaminated, robust = "auto")

Detect Regime Changes in Time Series

Description

Main function for detecting regime changes (changepoints) in time series data. Supports multiple detection methods, both frequentist and Bayesian, and can operate in offline (retrospective) or online (sequential) modes.

Usage

detect_regimes(
  data,
  method = c("pelt", "bocpd", "cusum", "binseg", "wbs", "shiryaev", "ensemble"),
  type = c("both", "mean", "variance", "trend", "distribution"),
  mode = c("offline", "online"),
  n_changepoints = "multiple",
  penalty = "BIC",
  min_segment = 2,
  prior = NULL,
  hazard = NULL,
  threshold = NULL,
  uncertainty = TRUE,
  bootstrap_reps = 200,
  ...
)

Arguments

Value

An object of class "regime_result" containing:

• changepoints: Vector of detected changepoint locations

• n_changepoints: Number of changepoints detected

• segments: List of segment information (start, end, parameters)

• confidence_intervals: Confidence intervals for changepoint locations

• existence_probability: Probability that each changepoint exists

• posterior: Posterior distribution (Bayesian methods)

• information_criterion: BIC/AIC values

• method: Method used

• call: The function call

Examples

set.seed(123)
data <- c(rnorm(100, mean = 0), rnorm(100, mean = 3))


result <- detect_regimes(data)
print(result)
plot(result)


result <- detect_regimes(data, method = "bocpd",
                         prior = normal_gamma())


result <- detect_regimes(data, method = "cusum",
                         mode = "online", threshold = 5)

Economic Cycles Dataset

Description

A simulated time series representing an economic indicator with regime changes corresponding to expansion, recession, recovery, and stable growth periods.

Usage

economic_cycles

Format

A time series object of length 500 with attributes:

true_changepoints

Vector of true changepoint locations: c(120, 250, 380)

description

Description of the dataset

regimes

Names of the regimes: Expansion, Recession, Recovery, Stable Growth

Details

The data simulates monthly economic data from 2000-2041 with four distinct regimes:

• Regime 1 (1-120): Expansion - positive trend, low volatility

• Regime 2 (121-250): Recession - negative trend, high volatility

• Regime 3 (251-380): Recovery - positive trend, medium volatility

• Regime 4 (381-500): Stable Growth - low trend, low volatility

Source

Simulated data for package examples

Examples

data(economic_cycles)
result <- detect_regimes(economic_cycles, method = "pelt")
plot(result)

# Compare with true changepoints
true_cps <- attr(economic_cycles, "true_changepoints")
evaluate(result, true_changepoints = true_cps)

E-Divisive Changepoint Detection

Description

Nonparametric changepoint detection using energy statistics. Can detect changes in any aspect of the distribution.

Usage

edivisive_detect(data, min_segment = 5, alpha = 0.05, n_perm = 199)

Arguments

Value

List with changepoints and test statistics

References

Matteson, D. S., and James, N. A. (2014). A nonparametric approach for multiple change point analysis of multivariate data. Journal of the American Statistical Association, 109(505), 334-345.

Ensemble Deep Learning Detection

Description

Combines multiple deep learning methods for robust detection.

Usage

ensemble_dl_detect(
  data,
  methods = c("autoencoder", "tcn", "transformer"),
  min_agreement = 2,
  ...
)

Arguments

Value

List with consensus changepoints and individual results

Evaluate Changepoint Detection Results

Description

Comprehensive evaluation of detection results against known ground truth. Computes multiple metrics for localization, segmentation, and detection.

Usage

evaluate(result, true_changepoints, n = NULL, tolerance = 5)

Arguments

Value

A list of class "regime_evaluation" containing all metrics

Examples

true_cp <- c(50, 100)
data <- c(rnorm(50), rnorm(50, mean = 2), rnorm(50))

result <- detect_regimes(data)
evaluation <- evaluate(result, true_cp)
print(evaluation)

Evaluation Metrics for Changepoint Detection

Description

Functions for evaluating changepoint detection results against ground truth. Includes metrics for localization accuracy, segmentation quality, and detection performance.

F1 Score with Tolerance

Description

Harmonic mean of precision and recall.

Usage

f1_score(detected, true_cp, tolerance = 5)

Arguments

Value

F1 score (0 to 1)

Examples

f1_score(c(48, 102, 150), c(50, 100), tolerance = 5)

FPOP - Functional Pruning Optimal Partitioning

Description

More efficient than PELT for certain data types by maintaining piecewise quadratic cost functions instead of just minimum values. Achieves O(n) complexity in practice.

Usage

fpop_detect(data, penalty = "bic", min_segment = 2, cost_type = "mean")

Arguments

Value

List with:

References

Maidstone, R., Hocking, T., Rigaill, G., and Fearnhead, P. (2017). On optimal multiple changepoint algorithms for large data. Statistics and Computing, 27(2), 519-533.

Examples

data <- c(rnorm(100, 0), rnorm(100, 3), rnorm(100, 1))
result <- fpop_detect(data, penalty = "bic")
print(result$changepoints)

Geometric Hazard Prior

Description

Creates a geometric (constant hazard) prior for the changepoint process. This implies that the probability of a changepoint at each time step is constant and independent.

Usage

geometric_hazard(lambda = 0.01)

Arguments

Details

Under a geometric hazard, the expected run length (time between changepoints) is 1/lambda. For example, lambda = 0.01 expects a changepoint every 100 observations on average.

Value

An object of class "hazard_prior"

Examples

hazard <- geometric_hazard(lambda = 0.01)

hazard <- geometric_hazard(lambda = 0.1)

Hausdorff Distance Between Changepoint Sets

Description

Computes the Hausdorff distance between detected and true changepoints. The Hausdorff distance is the maximum distance from a point in one set to the nearest point in the other set.

Usage

hausdorff_distance(detected, true_cp)

Arguments

Value

Hausdorff distance (non-negative number)

Examples

hausdorff_distance(c(48, 102), c(50, 100))

Industrial Sensor Dataset

Description

A simulated time series from a manufacturing process with abrupt changes representing process shifts and equipment states.

Usage

industrial_sensor

Format

A time series object of length 600 with attributes:

true_changepoints

Vector of true changepoint locations: c(150, 300, 450)

description

Description of the dataset

regimes

Names of the regimes: Normal, Drift, Malfunction, Corrected

Details

The data simulates per-minute sensor measurements with four distinct regimes:

• Regime 1 (1-150): Normal operation - mean=50, sd=2

• Regime 2 (151-300): Process drift - mean=55, sd=2.5

• Regime 3 (301-450): Equipment malfunction - mean=45, sd=5

• Regime 4 (451-600): Corrected operation - mean=50, sd=1.5

The data includes autocorrelation typical of industrial processes.

Source

Simulated data for package examples

Examples

data(industrial_sensor)
result <- detect_regimes(industrial_sensor, method = "bocpd")
plot(result, type = "segments")

Initialize Julia Backend

Description

Explicitly initialize the Julia backend for improved performance. This loads the RegimeChangeJulia module and makes Julia functions available.

Usage

init_julia(force = FALSE, num_threads = NULL)

Arguments

Value

Logical indicating success (invisibly).

Examples

## Not run: 
init_julia()
julia_available()
result <- detect_regimes(data, method = "pelt")

## End(Not run)

Inverse-Gamma Prior for Variance Only

Description

Creates an Inverse-Gamma prior for detecting changes in variance with known mean.

Usage

inverse_gamma_var(alpha0 = 1, beta0 = 1, known_mean = 0)

Arguments

Value

An object of class "regime_prior"

Check if Julia Backend is Available

Description

Check if Julia Backend is Available

Usage

julia_available()

Value

Logical indicating if Julia is available and initialized.

Get Julia Backend Status

Description

Get Julia Backend Status

Usage

julia_status()

Value

List with Julia status information.

Kernel-based Changepoint Detection

Description

Detects changepoints using Maximum Mean Discrepancy (MMD) in a Reproducing Kernel Hilbert Space. This nonparametric approach can detect changes in any aspect of the distribution.

Usage

kernel_cpd_detect(
  data,
  penalty = "bic",
  kernel = "rbf",
  bandwidth = 0,
  min_segment = 5
)

Arguments

Value

List with changepoints and kernel statistics

References

Arlot, S., Celisse, A., and Harchaoui, Z. (2019). A kernel multiple change-point algorithm via model selection. Journal of Machine Learning Research, 20(162), 1-56.

Mean Absolute Error for Changepoints

Description

Computes mean absolute error between matched changepoints. Unmatched changepoints are ignored.

Usage

mean_absolute_error(detected, true_cp)

Arguments

Value

Mean absolute error

Negative Binomial Hazard Prior

Description

Creates a negative binomial hazard prior, which allows for more flexibility in the distribution of run lengths.

Usage

negbin_hazard(r = 1, p = 0.01)

Arguments

Value

An object of class "hazard_prior"

Normal-Gamma Prior for Unknown Mean and Variance

Description

Creates a Normal-Gamma prior specification for data with unknown mean and variance. This is conjugate for normal observations.

Usage

normal_gamma(mu0 = 0, kappa0 = 1, alpha0 = 1, beta0 = 1)

Arguments

Details

The Normal-Gamma prior places a joint distribution on (mu, tau) where tau = 1/sigma^2:

tau ~ Gamma(alpha0, beta0) mu | tau ~ Normal(mu0, 1/(kappa0 * tau))

The prior mean of mu is mu0, and the prior mean of sigma^2 is beta0/(alpha0-1) for alpha0 > 1.

Value

An object of class "regime_prior"

Examples

prior <- normal_gamma()

prior <- normal_gamma(mu0 = 0, kappa0 = 10, alpha0 = 3, beta0 = 2)

Normal Prior for Unknown Mean with Known Variance

Description

Creates a Normal prior specification for data with unknown mean but known variance.

Usage

normal_known_var(mu0 = 0, sigma0 = 1, known_var = 1)

Arguments

Value

An object of class "regime_prior"

Examples

prior <- normal_known_var(mu0 = 0, sigma0 = 1, known_var = 1)

Normal-Wishart Prior for Multivariate Data

Description

Creates a Normal-Wishart prior for multivariate normal data with unknown mean vector and covariance matrix.

Usage

normal_wishart(mu0, kappa0 = 1, nu0 = NULL, Psi0 = NULL)

Arguments

Value

An object of class "regime_prior"

Examples

prior <- normal_wishart(
  mu0 = c(0, 0),
  kappa0 = 1,
  nu0 = 3,
  Psi0 = diag(2)
)

NOT - Narrowest-Over-Threshold Changepoint Detection

Description

Detects changepoints by finding the narrowest intervals that contain significant changes, providing excellent localization.

Usage

not_detect(
  data,
  threshold = NULL,
  min_segment = 5,
  contrast_type = "pcwsConst"
)

Arguments

Value

List with changepoints and interval information

References

Baranowski, R., Chen, Y., and Fryzlewicz, P. (2019). Narrowest-over- threshold detection of multiple change points and change-point-like features. Journal of the Royal Statistical Society Series B, 81(3), 649-672.

Detect changepoints using PELT algorithm

Description

Wrapper for detect_pelt for compatibility.

Usage

pelt(
  data,
  type = "both",
  penalty = "MBIC",
  min_segment = 2,
  robust = FALSE,
  correct_ar = FALSE,
  merge_close = NULL,
  ...
)

Arguments

Value

List with changepoints and diagnostics

See Also

detect_pelt

Plot Regime Change Detection Results

Description

Create visualizations of changepoint detection results.

Usage

## S3 method for class 'regime_result'
plot(
  x,
  type = c("data", "segments", "posterior", "diagnostic", "runlength"),
  show_ci = TRUE,
  show_segments = TRUE,
  title = NULL,
  ...
)

Arguments

Value

A ggplot2 object

Examples

data <- c(rnorm(100), rnorm(100, mean = 2))
result <- detect_regimes(data)
plot(result)
plot(result, type = "segments")

Plot Multiple Results for Comparison

Description

Creates a faceted plot comparing results from different methods.

Usage

plot_compare(..., names = NULL)

Arguments

Value

A ggplot object

Create Interactive Plot

Description

Creates an interactive version of the changepoint plot using plotly.

Usage

plot_interactive(result, ...)

Arguments

Value

A plotly object

Create Multi-Panel Summary Plot

Description

Creates a comprehensive multi-panel visualization of detection results.

Usage

plot_summary(result, ...)

Arguments

Value

A combined ggplot object

Gamma-Poisson Prior for Count Data

Description

Creates a Gamma prior specification for Poisson-distributed count data.

Usage

poisson_gamma(alpha0 = 1, beta0 = 1)

Arguments

Value

An object of class "regime_prior"

Examples

prior <- poisson_gamma(alpha0 = 1, beta0 = 1)

Precision Score with Tolerance

Description

Proportion of detected changepoints that are within tolerance of a true one.

Usage

precision_score(detected, true_cp, tolerance = 5)

Arguments

Value

Precision (0 to 1)

Rand Index for Segmentation

Description

Measures agreement between detected and true segmentations. Based on pairwise comparisons of whether points are in the same segment.

Usage

rand_index(detected, true_cp, n)

Arguments

Value

Rand Index (0 to 1)

Recall Score with Tolerance

Description

Proportion of true changepoints that are matched by a detection.

Usage

recall_score(detected, true_cp, tolerance = 5)

Arguments

Value

Recall (0 to 1)

Create Online Regime Detector

Description

Creates a detector object for online (sequential) changepoint detection. The detector maintains state and can be updated incrementally as new observations arrive.

Usage

regime_detector(
  method = c("bocpd", "cusum", "shiryaev"),
  prior = NULL,
  hazard = NULL,
  threshold = NULL,
  ...
)

Arguments

Value

An object of class "regime_detector"

Examples

detector <- regime_detector(method = "bocpd",
                            prior = normal_gamma(),
                            threshold = 0.5)


for (x in rnorm(100)) {
  detector <- update(detector, x)
  if (detector$last_result$alarm) {
    message("Change detected at observation ", detector$last_result$t)
    detector <- reset(detector)
  }
}

Reset a detector to its initial state

Description

Reset a detector to its initial state

Usage

reset(object, ...)

## S3 method for class 'regime_detector'
reset(object, ...)

Arguments

Value

The reset detector object

Methods (by class)

• reset(regime_detector): Reset method for regime_detector

RMSE for Changepoints

Description

RMSE for Changepoints

Usage

rmse_changepoints(detected, true_cp)

Arguments

Value

Root mean squared error

Shiryaev-Roberts Changepoint Detection

Description

Implements the Shiryaev-Roberts procedure, which is asymptotically optimal for detecting changes with minimal detection delay.

Usage

shiryaev_roberts(
  data,
  type = "mean",
  prior = NULL,
  hazard = NULL,
  threshold = 100,
  mu0 = NULL,
  mu1 = NULL,
  sigma = NULL,
  ...
)

Arguments

Value

List with changepoints and statistics

References

Shiryaev, A. N. (1963). On Optimum Methods in Quickest Detection Problems. Theory of Probability and Its Applications.

Examples

data <- c(rnorm(100), rnorm(100, mean = 1))
result <- shiryaev_roberts(data)

Simulated Changepoints Benchmark Dataset

Description

A collection of simulated datasets with known changepoints for benchmarking changepoint detection methods.

Usage

simulated_changepoints

Format

A list with multiple scenarios:

single_mean

Single mean change from 0 to 3

single_variance

Single variance change from 1 to 9

multiple_mean

Three mean changes

gradual

Gradual trend change (challenging)

small_change

Small mean change of 0.5 SD (challenging)

close_changepoints

Three closely spaced changepoints (challenging)

heavy_tailed

Mean change with t-distributed noise

multivariate

Bivariate mean and covariance change

Each scenario contains:

• data: The time series data

• true_changepoints: Vector of true changepoint locations

• type: Type of change (mean, variance, etc.)

• description: Description of the scenario

Source

Simulated data for package benchmarking

Examples

data(simulated_changepoints)

# Run benchmark on single mean scenario
scenario <- simulated_changepoints$single_mean
result <- detect_regimes(scenario$data, method = "pelt")
evaluate(result, true_changepoints = scenario$true_changepoints)

# Compare multiple methods
comparison <- compare_methods(
  data = scenario$data,
  methods = c("pelt", "bocpd", "binseg"),
  true_changepoints = scenario$true_changepoints
)
print(comparison)

Sparse Projection Changepoint Detection

Description

Detects changepoints in high-dimensional data using random sparse projections to reduce dimensionality while preserving changepoint structure.

Usage

sparse_projection_cpd(
  data,
  n_projections = 10,
  penalty = "bic",
  min_segment = 5,
  sparsity = 0.3
)

Arguments

Value

List with changepoints and projection information

References

Wang, D. and Bhattacharjee, M. (2021). High-dimensional changepoint estimation via sparse projection. arXiv preprint.

TCN-based Changepoint Detection

Description

Uses Temporal Convolutional Networks with dilated causal convolutions for sequence-to-sequence changepoint prediction.

Usage

tcn_detect(
  data,
  true_changepoints = NULL,
  window_size = 64,
  n_filters = 64,
  kernel_size = 3,
  dilations = c(1, 2, 4, 8, 16),
  dropout = 0.2,
  epochs = 50,
  threshold = 0.5,
  verbose = FALSE
)

Arguments

Value

List with changepoints, probabilities, and model

References

Bai, S., Kolter, J. Z., and Koltun, V. (2018). An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling

Transformer-based Changepoint Detection

Description

Implements a transformer architecture inspired by TCDformer for time series changepoint detection using self-attention mechanisms.

Usage

transformer_detect(
  data,
  true_changepoints = NULL,
  window_size = 128,
  d_model = 64,
  n_heads = 4,
  n_layers = 2,
  d_ff = 256,
  dropout = 0.1,
  epochs = 50,
  threshold = 0.5,
  verbose = FALSE
)

Arguments

Value

List with changepoints, attention weights, and model

References

Wu, H., et al. (2023). TimesNet: Temporal 2D-Variation Modeling

Zhou, H., et al. (2021). Informer: Efficient Transformer for Long Sequence Time-Series Forecasting

Update Online Detector with New Observation

Description

Update Online Detector with New Observation

Usage

## S3 method for class 'regime_detector'
update(object, x, ...)

Arguments

Value

The updated regime_detector object with results in $last_result

Well Log Dataset

Description

A simulated well-log porosity dataset with abrupt lithology changes typical of geological formations.

Usage

well_log

Format

A numeric vector of length 1000 with attributes:

true_changepoints

Vector of true changepoint locations: c(200, 350, 500, 700, 850)

description

Description of the dataset

lithologies

Names of lithology units

Details

The data simulates porosity measurements from a well log with six distinct lithological units:

• Unit 1 (1-200): Sandstone - porosity ~15%

• Unit 2 (201-350): Shale - porosity ~8%

• Unit 3 (351-500): Limestone - porosity ~20%

• Unit 4 (501-700): Sandstone - porosity ~12%

• Unit 5 (701-850): Dense Shale - porosity ~5%

• Unit 6 (851-1000): Sandstone - porosity ~18%

Source

Simulated data based on typical well-log characteristics

Examples

data(well_log)
result <- detect_regimes(well_log, method = "pelt", min_segment = 50)
plot(result, type = "segments")

# Compare with true lithology boundaries
true_cps <- attr(well_log, "true_changepoints")
evaluate(result, true_changepoints = true_cps)

Wild Binary Segmentation

Description

Detects multiple changepoints using the Wild Binary Segmentation algorithm. Uses random intervals to improve detection in long time series.

Usage

wild_binary_segmentation(
  data,
  type = "both",
  penalty = "BIC",
  min_segment = 2,
  n_changepoints = "multiple",
  M = 5000,
  threshold = NULL,
  ...
)

Arguments

Value

A list with:

References

Fryzlewicz, P. (2014). Wild Binary Segmentation for multiple change-point detection. Annals of Statistics, 42(6), 2243-2281.

Examples

data <- c(rnorm(100), rnorm(100, mean = 2), rnorm(100))
result <- wild_binary_segmentation(data)