Smoothing Long-Memory Time Series

The nonparametric trend and its derivatives in equidistant time series (TS) with long-memory errors can be estimated. The estimation is conducted via local polynomial regression using an automatically selected bandwidth obtained by a built-in iterative plug-in algorithm or a bandwidth fixed by the user. The smoothing methods of the package are described in Letmathe, S., Beran, J. and Feng, Y., (2023) .

Multivariate Time Series Data Imputation

This is an EM algorithm based method for imputation of missing values in multivariate normal time series. The imputation algorithm accounts for both spatial and temporal correlation structures. Temporal patterns can be modeled using an ARIMA(p,d,q), optionally with seasonal components, a non-parametric cubic spline or generalized additive models with exogenous covariates. This algorithm is specially tailored for climate data with missing measurements from several monitors along a given region.

Manipulate Time Series of Climate Reconstructions

Methods to easily extract and manipulate climate reconstructions for ecological and anthropological analyses, as described in Leonardi et al. (2023) . The package includes datasets of palaeoclimate reconstructions, present observations, and future projections from multiple climate models.

Mixture Hidden Markov Models for Social Sequence Data and Other Multivariate, Multichannel Categorical Time Series

Designed for estimating variants of hidden (latent) Markov models (HMMs), mixture HMMs, and non-homogeneous HMMs (NHMMs) for social sequence data and other categorical time series. Special cases include feedback-augmented NHMMs, Markov models without latent layer, mixture Markov models, and latent class models. The package supports models for one or multiple subjects with one or multiple parallel sequences (channels). External covariates can be added to explain cluster membership in mixture models as well as initial, transition and emission probabilities in NHMMs. The package provides functions for evaluating and comparing models, as well as functions for visualizing of multichannel sequence data and HMMs. For NHMMs, methods for computing average causal effects and marginal state and emission probabilities are available. Models are estimated using maximum likelihood via the EM algorithm or direct numerical maximization with analytical gradients. Documentation is available via several vignettes, and Helske and Helske (2019, ). For methodology behind the NHMMs, see Helske (2025, ).

Inferring Causal Effects using Bayesian Structural Time-Series Models

Implements a Bayesian approach to causal impact estimation in time series, as described in Brodersen et al. (2015) . See the package documentation on GitHub < https://google.github.io/CausalImpact/> to get started.

Rmetrics - Nonlinear and Chaotic Time Series Modelling

Provides a collection of functions for testing various aspects of univariate time series including independence and neglected nonlinearities. Further provides functions to investigate the chaotic behavior of time series processes and to simulate different types of chaotic time series maps.

Rmetrics - Autoregressive Conditional Heteroskedastic Modelling

Analyze and model heteroskedastic behavior in financial time series.

Nonlinear Time Series Models with Regime Switching

Implements nonlinear autoregressive (AR) time series models. For univariate series, a non-parametric approach is available through additive nonlinear AR. Parametric modeling and testing for regime switching dynamics is available when the transition is either direct (TAR: threshold AR) or smooth (STAR: smooth transition AR, LSTAR). For multivariate series, one can estimate a range of TVAR or threshold cointegration TVECM models with two or three regimes. Tests can be conducted for TVAR as well as for TVECM (Hansen and Seo 2002 and Seo 2006).

Imputation of Time Series Based on Dynamic Time Warping

Functions to impute large gaps within time series based on Dynamic Time Warping methods. It contains all required functions to create large missing consecutive values within time series and to fill them, according to the paper Phan et al. (2017), . Performance criteria are added to compare similarity between two signals (query and reference).

Dimension Reduction Methods for Multivariate Time Series

Estimates VAR and VARX models with Structured Penalties.