Superfast Likelihood Inference for Stationary Gaussian Time Series

Likelihood evaluations for stationary Gaussian time series are typically obtained via the Durbin-Levinson algorithm, which scales as O(n^2) in the number of time series observations. This package provides a "superfast" O(n log^2 n) algorithm written in C++, crossing over with Durbin-Levinson around n = 300. Efficient implementations of the score and Hessian functions are also provided, leading to superfast versions of inference algorithms such as Newton-Raphson and Hamiltonian Monte Carlo. The C++ code provides a Toeplitz matrix class packaged as a header-only library, to simplify low-level usage in other packages and outside of R.

Two Functions for Generalized SARIMA Time Series Simulation

Write SARIMA models in (finite) AR representation and simulate generalized multiplicative seasonal autoregressive moving average (time) series with Normal / Gaussian, Poisson or negative binomial distribution. The methodology of this method is described in Briet OJT, Amerasinghe PH, and Vounatsou P (2013) .

Methods for Handling and Analyzing Time Series of Satellite Images

Provides functions and methods for: splitting large raster objects into smaller chunks, transferring images from a binary format into raster layers, transferring raster layers into an 'RData' file, calculating the maximum gap (amount of consecutive missing values) of a numeric vector, and fitting harmonic regression models to periodic time series. The homoscedastic harmonic regression model is based on G. Roerink, M. Menenti and W. Verhoef (2000) .

Time Series Regression

Time series regression. The dyn class interfaces ts, irts(), zoo() and zooreg() time series classes to lm(), glm(), loess(), quantreg::rq(), MASS::rlm(), MCMCpack::MCMCregress(), quantreg::rq(), randomForest::randomForest() and other regression functions allowing those functions to be used with time series including specifications that may contain lags, diffs and missing values.

Archaeological Time Series

A toolkit for archaeological time series and time intervals. This package provides a system of classes and methods to represent and work with archaeological time series and time intervals. Dates are represented as "rata die" and can be converted to (virtually) any calendar defined by Reingold and Dershowitz (2018) . This packages offers a simple API that can be used by other specialized packages.

Time Series Analysis

Analyzes autocorrelation and partial autocorrelation using surrogate methods and bootstrapping, and computes the acceleration constants for the vectorized moving block bootstrap provided by this package. It generates percentile, bias-corrected, and accelerated intervals and estimates partial autocorrelations using Durbin-Levinson. This package calculates the autocorrelation power spectrum, computes cross-correlations between two time series, computes bandwidth for any time series, and performs autocorrelation frequency analysis. It also calculates the periodicity of a time series.

'data.table' Time-Series

High-frequency time-series support via 'nanotime' and 'data.table'.

Time Series Plot

A fast and elegant time series visualization package. In addition to the standard R plot types, this package supports candle sticks, open-high-low-close, and volume plots. Useful for visualizing any time series data, e.g., stock prices and technical indicators.

Time Series Methods

Generic methods for use in a time series probabilistic framework, allowing for a common calling convention across packages. Additional methods for time series prediction ensembles and probabilistic plotting of predictions is included. A more detailed description is available at < https://www.nopredict.com/packages/tsmethods> which shows the currently implemented methods in the 'tsmodels' framework.

Time Series Representations

Methods for representations (i.e. dimensionality reduction, preprocessing, feature extraction) of time series to help more accurate and effective time series data mining. Non-data adaptive, data adaptive, model-based and data dictated (clipped) representation methods are implemented. Also various normalisation methods (min-max, z-score, Box-Cox, Yeo-Johnson), and forecasting accuracy measures are implemented.