| Title: | Time Series Forecasting with uncertainty quantification |
|---|---|
| Description: | Univariate and multivariate time series forecasting with uncertainty quantification. |
| Authors: | T. Moudiki |
| Maintainer: | T. Moudiki <[email protected]> |
| License: | BSD_3_clause Clear + file LICENSE |
| Version: | 0.14.3 |
| Built: | 2024-11-23 13:31:45 UTC |
| Source: | https://github.com/Techtonique/ahead |
ARMA(1, 1)-GARCH(1, 1) forecasting (with simulation)
armagarchf( y, h = 5, level = 95, B = 250, cl = 1L, dist = c("student", "gaussian"), seed = 123 )armagarchf( y, h = 5, level = 95, B = 250, cl = 1L, dist = c("student", "gaussian"), seed = 123 )
y |
a univariate time series |
h |
number of periods for forecasting |
level |
confidence level for prediction intervals |
B |
number of simulations for |
cl |
an integer; the number of clusters for parallel execution |
dist |
distribution of innovations ("student" or "gaussian") |
seed |
reproducibility seed |
An object of class "forecast"; a list containing the following elements:
model |
A list containing information about the fitted model |
method |
The name of the forecasting method as a character string |
mean |
Point forecasts for the time series |
lower |
Lower bound for prediction interval |
upper |
Upper bound for prediction interval |
x |
The original time series |
sims |
Simulations of ARMA(1, 1)-GARCH(1, 1) |
T. Moudiki
y <- datasets::EuStockMarkets[ , "DAX"] log_returns <- ts(log(y[-1]/y[-length(y)])) # require(forecast) # z <- ahead::armagarchf(y=log_returns, h=200) # plot(z)y <- datasets::EuStockMarkets[ , "DAX"] log_returns <- ts(log(y[-1]/y[-length(y)])) # require(forecast) # z <- ahead::armagarchf(y=log_returns, h=200) # plot(z)
Basic forecasting functions for multivariate time series
basicf( y, h = 5, level = 95, method = c("mean", "median", "rw"), type_pi = c("gaussian", "bootstrap", "blockbootstrap", "movingblockbootstrap"), block_length = NULL, seed = 1, B = 100, show_progress = TRUE )basicf( y, h = 5, level = 95, method = c("mean", "median", "rw"), type_pi = c("gaussian", "bootstrap", "blockbootstrap", "movingblockbootstrap"), block_length = NULL, seed = 1, B = 100, show_progress = TRUE )
y |
A multivariate time series of class |
h |
Forecasting horizon |
level |
Confidence level for prediction intervals |
method |
forecasting method, either "mean", "median", or random walk ("rw") |
type_pi |
type of prediction interval currently, "gaussian", "bootstrap", "blockbootstrap" or "movingblockbootstrap" |
block_length |
length of block for (circular) "blockbootstrap" or "movingblockbootstrap" |
seed |
reproducibility seed for |
B |
Number of bootstrap replications for |
show_progress |
A boolean; show progress bar for bootstrapping? Default is TRUE. |
An object of class "mtsforecast"; a list containing the following elements:
method |
The name of the forecasting method as a character string |
mean |
Point forecasts for the time series |
lower |
Lower bound for prediction interval |
upper |
Upper bound for prediction interval |
sims |
Model simulations for bootstrapping (basic, or block) |
x |
The original time series |
residuals |
Residuals from the fitted model |
coefficients |
Regression coefficients for |
require(fpp) res <- ahead::basicf(fpp::insurance, h=10) par(mfrow=c(1, 2)) plot(res, "TV.advert") plot(res, "Quotes") res <- ahead::basicf(fpp::insurance, method="rw", h=10) par(mfrow=c(1, 2)) plot(res, "TV.advert") plot(res, "Quotes") # block bootstrap res3 <- ahead::basicf(fpp::insurance, h=10, type_pi = "bootstrap", B=10) res5 <- ahead::basicf(fpp::insurance, h=10, type_pi = "blockbootstrap", B=10, block_length = 4) print(res3$sims[[2]]) print(res5$sims[[2]]) par(mfrow=c(2, 2)) plot(res3, "Quotes") plot(res3, "TV.advert") plot(res5, "Quotes") plot(res5, "TV.advert") # moving block bootstrap res6 <- ahead::basicf(fpp::insurance, h=10, type_pi = "movingblockbootstrap", B=10, block_length = 4, method = "rw") par(mfrow=c(1, 2)) plot(res6, "Quotes") plot(res6, "TV.advert")require(fpp) res <- ahead::basicf(fpp::insurance, h=10) par(mfrow=c(1, 2)) plot(res, "TV.advert") plot(res, "Quotes") res <- ahead::basicf(fpp::insurance, method="rw", h=10) par(mfrow=c(1, 2)) plot(res, "TV.advert") plot(res, "Quotes") # block bootstrap res3 <- ahead::basicf(fpp::insurance, h=10, type_pi = "bootstrap", B=10) res5 <- ahead::basicf(fpp::insurance, h=10, type_pi = "blockbootstrap", B=10, block_length = 4) print(res3$sims[[2]]) print(res5$sims[[2]]) par(mfrow=c(2, 2)) plot(res3, "Quotes") plot(res3, "TV.advert") plot(res5, "Quotes") plot(res5, "TV.advert") # moving block bootstrap res6 <- ahead::basicf(fpp::insurance, h=10, type_pi = "movingblockbootstrap", B=10, block_length = 4, method = "rw") par(mfrow=c(1, 2)) plot(res6, "Quotes") plot(res6, "TV.advert")
Computes forecast combination weights using GLMNET Regression (OLS) regression.
comb_GLMNET(x, custom_error = NULL)comb_GLMNET(x, custom_error = NULL)
x |
An object of class 'foreccomb'. Contains training set (actual values + matrix of model forecasts) and optionally a test set. |
The function integrates the GLMNET Regression forecast combination implementation of the ForecastCombinations package into ForecastComb.
The results are stored in an object of class 'ForecastComb::foreccomb_res', for which separate plot and summary functions are provided.
Returns an object of class ForecastComb::foreccomb_res with the following components:
Method |
Returns the best-fit forecast combination method. |
Models |
Returns the individual input models that were used for the forecast combinations. |
Weights |
Returns the combination weights obtained by applying the combination method to the training set. |
Intercept |
Returns the intercept of the linear regression. |
Fitted |
Returns the fitted values of the combination method for the training set. |
Accuracy_Train |
Returns range of summary measures of the forecast accuracy for the training set. |
Forecasts_Test |
Returns forecasts produced by the combination method for the test set. Only returned if input included a forecast matrix for the test set. |
Accuracy_Test |
Returns range of summary measures of the forecast accuracy for the test set. Only returned if input included a forecast matrix and a vector of actual values for the test set. |
Input_Data |
Returns the data forwarded to the method. |
Forecast_comb,
foreccomb,
plot.ForecastComb::foreccomb_res,
summary.ForecastComb::foreccomb_res,
accuracy
library(ForecastComb) data(electricity) print(head(electricity)) forecasting_methods <- colnames(electricity)[1:5] train_obs <- electricity[1:84, "Actual"] train_pred <- electricity[1:84, forecasting_methods] test_obs <- electricity[85:123, "Actual"] test_pred <- electricity[85:123, forecasting_methods] data <- ForecastComb::foreccomb(train_obs, train_pred, test_obs, test_pred) # obj <- ahead::comb_GLMNET(data))library(ForecastComb) data(electricity) print(head(electricity)) forecasting_methods <- colnames(electricity)[1:5] train_obs <- electricity[1:84, "Actual"] train_pred <- electricity[1:84, forecasting_methods] test_obs <- electricity[85:123, "Actual"] test_pred <- electricity[85:123, forecasting_methods] data <- ForecastComb::foreccomb(train_obs, train_pred, test_obs, test_pred) # obj <- ahead::comb_GLMNET(data))
Computes forecast combination weights using ordinary least squares (OLS) regression.
comb_OLS(x, custom_error = NULL)comb_OLS(x, custom_error = NULL)
x |
An object of class 'foreccomb'. Contains training set (actual values + matrix of model forecasts) and optionally a test set. |
The function integrates the ordinary least squares (OLS) forecast combination implementation of the ForecastCombinations package into ForecastComb.
The OLS combination method (Granger and Ramanathan (1984)) uses ordinary least squares to
estimate the weights, , as well as an intercept, , for the combination of
the forecasts.
Suppose that there are not perfectly collinear predictors ,
then the forecast combination for one data point can be represented as:
An appealing feature of the method is its bias correction through the intercept – even if one or more of the individual
predictors are biased, the resulting combined forecast is unbiased. A disadvantage of the method is that it places no
restriction on the combination weights (i.e., they do not add up to 1 and can be negative), which can make interpretation
hard. Another issue, documented in Nowotarski et al. (2014), is the method's unstable behavior
when predictors are highly correlated (which is the norm in forecast combination): Minor fluctuations in the sample
can cause major shifts of the coefficient vector (‘bouncing betas’) – often causing poor out-of-sample performance.
This issue is addressed by the comb_LAD method that is more robust to outliers.
The results are stored in an object of class 'ForecastComb::foreccomb_res', for which separate plot and summary functions are provided.
Returns an object of class ForecastComb::foreccomb_res with the following components:
Method |
Returns the best-fit forecast combination method. |
Models |
Returns the individual input models that were used for the forecast combinations. |
Weights |
Returns the combination weights obtained by applying the combination method to the training set. |
Intercept |
Returns the intercept of the linear regression. |
Fitted |
Returns the fitted values of the combination method for the training set. |
Accuracy_Train |
Returns range of summary measures of the forecast accuracy for the training set. |
Forecasts_Test |
Returns forecasts produced by the combination method for the test set. Only returned if input included a forecast matrix for the test set. |
Accuracy_Test |
Returns range of summary measures of the forecast accuracy for the test set. Only returned if input included a forecast matrix and a vector of actual values for the test set. |
Input_Data |
Returns the data forwarded to the method. |
Forecast_comb,
foreccomb,
plot.ForecastComb::foreccomb_res,
summary.ForecastComb::foreccomb_res,
accuracy
obs <- rnorm(100) preds <- matrix(rnorm(1000, 1), 100, 10) train_o<-obs[1:80] train_p<-preds[1:80,] test_o<-obs[81:100] test_p<-preds[81:100,] data<-ForecastComb::foreccomb(train_o, train_p, test_o, test_p) ahead::comb_OLS(data)obs <- rnorm(100) preds <- matrix(rnorm(1000, 1), 100, 10) train_o<-obs[1:80] train_p<-preds[1:80,] test_o<-obs[81:100] test_p<-preds[81:100,] data<-ForecastComb::foreccomb(train_o, train_p, test_o, test_p) ahead::comb_OLS(data)
Computes forecast combination weights using Ridge Regression (OLS) regression.
comb_Ridge(x, custom_error = NULL)comb_Ridge(x, custom_error = NULL)
x |
An object of class 'foreccomb'. Contains training set (actual values + matrix of model forecasts) and optionally a test set. |
The function integrates the Ridge Regression forecast combination implementation of the ForecastCombinations package into ForecastComb.
The results are stored in an object of class 'ForecastComb::foreccomb_res', for which separate plot and summary functions are provided.
Returns an object of class ForecastComb::foreccomb_res with the following components:
Method |
Returns the best-fit forecast combination method. |
Models |
Returns the individual input models that were used for the forecast combinations. |
Weights |
Returns the combination weights obtained by applying the combination method to the training set. |
Intercept |
Returns the intercept of the linear regression. |
Fitted |
Returns the fitted values of the combination method for the training set. |
Accuracy_Train |
Returns range of summary measures of the forecast accuracy for the training set. |
Forecasts_Test |
Returns forecasts produced by the combination method for the test set. Only returned if input included a forecast matrix for the test set. |
Accuracy_Test |
Returns range of summary measures of the forecast accuracy for the test set. Only returned if input included a forecast matrix and a vector of actual values for the test set. |
Input_Data |
Returns the data forwarded to the method. |
Forecast_comb,
foreccomb,
plot.ForecastComb::foreccomb_res,
summary.ForecastComb::foreccomb_res,
accuracy
obs <- rnorm(100) preds <- matrix(rnorm(1000, 1), 100, 10) train_o<-obs[1:80] train_p<-preds[1:80,] test_o<-obs[81:100] test_p<-preds[81:100,] data<-ForecastComb::foreccomb(train_o, train_p, test_o, test_p) ahead::comb_Ridge(data)obs <- rnorm(100) preds <- matrix(rnorm(1000, 1), 100, 10) train_o<-obs[1:80] train_p<-preds[1:80,] test_o<-obs[81:100] test_p<-preds[81:100,] data<-ForecastComb::foreccomb(train_o, train_p, test_o, test_p) ahead::comb_Ridge(data)
This function allows to conformalize any forecasting function.
conformalize( FUN, y, h, level = 95, method = c("block-bootstrap", "surrogate", "kde", "bootstrap"), nsim = 100L, block_size = 5, seed = 123L, ... )conformalize( FUN, y, h, level = 95, method = c("block-bootstrap", "surrogate", "kde", "bootstrap"), nsim = 100L, block_size = 5, seed = 123L, ... )
FUN |
A forecasting function. |
y |
A time series ( |
h |
Forecasting horizon. |
level |
Confidence level. |
method |
Method to be used for conformalization (simulation of calibrated residuals). |
nsim |
Number of simulations. |
block_size |
Block size for block-bootstrap. |
seed |
Seed for reproducibility. |
... |
Additional arguments to be passed to the forecasting function. |
An object of class forecast.
y <- fdeaths h <- 25L obj <- conformalize(FUN=forecast::ets, y, h); plot(obj) obj <- conformalize(FUN=HoltWinters, y=y, h=h, seasonal = "mult"); plot(obj)y <- fdeaths h <- 25L obj <- conformalize(FUN=forecast::ets, y, h); plot(obj) obj <- conformalize(FUN=HoltWinters, y=y, h=h, seasonal = "mult"); plot(obj)
Create trend and seasonality features for univariate time series
createtrendseason(y)createtrendseason(y)
y |
a univariate time series |
a vector or matrix of features
y <- ts(rnorm(100), start = 1, frequency = 12) createtrendseason(y) createtrendseason(USAccDeaths)y <- ts(rnorm(100), start = 1, frequency = 12) createtrendseason(y) createtrendseason(USAccDeaths)
Adapted from forecast::nnetar, with alternative fitting functions (see examples)
dynrmf( y, h = 5, level = 95, fit_func = ahead::ridge, predict_func = predict, fit_params = NULL, type_pi = c("gaussian", "E", "A", "T"), xreg_fit = NULL, xreg_predict = NULL, ... )dynrmf( y, h = 5, level = 95, fit_func = ahead::ridge, predict_func = predict, fit_params = NULL, type_pi = c("gaussian", "E", "A", "T"), xreg_fit = NULL, xreg_predict = NULL, ... )
y |
A numeric vector or time series of class |
h |
Forecasting horizon |
level |
Confidence level for prediction intervals |
fit_func |
Fitting function (Statistical/ML model). Default is Ridge regression. |
predict_func |
Prediction function (Statistical/ML model) |
fit_params |
a list of additional parameters for the fitting function |
type_pi |
Type of prediction interval (currently "gaussian", ETS: "E", Arima: "A" or Theta: "T") |
xreg_fit |
Optionally, a vector or matrix of external regressors, which must have the same number of rows as y. Must be numeric. |
xreg_predict |
Future values of external regressor variables. |
... |
additional parameters |
a list; an object of class forecast.
The function summary is used to obtain and print a summary of the
results.
The generic accessor functions fitted.values and residuals
extract useful features.
T. Moudiki
Hyndman, R. J., & Athanasopoulos, G. (2018). Forecasting: principles and practice. OTexts.
Hyndman R, Athanasopoulos G, Bergmeir C, Caceres G, Chhay L,
O'Hara-Wild M, Petropoulos F, Razbash S, Wang E, Yasmeen F (2021).
forecast: Forecasting functions for time series and linear models. R
package version 8.14, <URL: https://pkg.robjhyndman.com/forecast/>.
# Example 0: with Ridge regression par(mfrow=c(3, 2)) plot(dynrmf(USAccDeaths, h=20, level=95)) plot(dynrmf(AirPassengers, h=20, level=95)) plot(dynrmf(lynx, h=20, level=95)) plot(dynrmf(WWWusage, h=20, level=95)) plot(dynrmf(Nile, h=20, level=95)) plot(dynrmf(fdeaths, h=20, level=95)) # Example 1: with Random Forest ## Not run: require(randomForest) par(mfrow=c(3, 2)) plot(dynrmf(USAccDeaths, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(AirPassengers, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(lynx, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(WWWusage, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(Nile, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) ## End(Not run) # Example 2: with SVM ## Not run: require(e1071) par(mfrow=c(2, 2)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = e1071::svm, fit_params = list(kernel = "linear"), predict_func = predict)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = e1071::svm, fit_params = list(kernel = "polynomial"), predict_func = predict)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = e1071::svm, fit_params = list(kernel = "radial"), predict_func = predict)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = e1071::svm, fit_params = list(kernel = "sigmoid"), predict_func = predict)) ## End(Not run)# Example 0: with Ridge regression par(mfrow=c(3, 2)) plot(dynrmf(USAccDeaths, h=20, level=95)) plot(dynrmf(AirPassengers, h=20, level=95)) plot(dynrmf(lynx, h=20, level=95)) plot(dynrmf(WWWusage, h=20, level=95)) plot(dynrmf(Nile, h=20, level=95)) plot(dynrmf(fdeaths, h=20, level=95)) # Example 1: with Random Forest ## Not run: require(randomForest) par(mfrow=c(3, 2)) plot(dynrmf(USAccDeaths, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(AirPassengers, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(lynx, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(WWWusage, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(Nile, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = randomForest::randomForest, fit_params = list(ntree = 50), predict_func = predict)) ## End(Not run) # Example 2: with SVM ## Not run: require(e1071) par(mfrow=c(2, 2)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = e1071::svm, fit_params = list(kernel = "linear"), predict_func = predict)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = e1071::svm, fit_params = list(kernel = "polynomial"), predict_func = predict)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = e1071::svm, fit_params = list(kernel = "radial"), predict_func = predict)) plot(dynrmf(fdeaths, h=20, level=95, fit_func = e1071::svm, fit_params = list(kernel = "sigmoid"), predict_func = predict)) ## End(Not run)
Combined ets, arima, and theta (eat) forecasting (uses forecast::ets,
forecast::auto.arima, forecast::thetaf)
eatf( y, h = 5, level = 95, method = c("EAT", "E", "A", "T"), weights = rep(1/3, 3), type_pi = c("gaussian", "E", "A", "T"), ... )eatf( y, h = 5, level = 95, method = c("EAT", "E", "A", "T"), weights = rep(1/3, 3), type_pi = c("gaussian", "E", "A", "T"), ... )
y |
a univariate time series |
h |
number of periods for forecasting |
level |
confidence level for prediction intervals |
method |
forecasting method: "E" for |
weights |
weights for each method, in method |
type_pi |
type of prediction interval: currently ETS: "E", Auto.Arima: "A" or Theta: "T" |
... |
additional parameters to be passed to |
ensemble forecasts obtained from forecast::ets,
forecast::auto.arima and forecast::theta (with weights)
An object of class "forecast"; a list containing the following elements:
model |
A list containing information about the fitted model |
method |
The name of the forecasting method as a character string |
mean |
Point forecasts for the time series |
lower |
Lower bound for prediction interval |
upper |
Upper bound for prediction interval |
x |
The original time series |
residuals |
Residuals from the fitted model |
T. Moudiki
Hyndman R, Athanasopoulos G, Bergmeir C, Caceres G, Chhay L,
O'Hara-Wild M, Petropoulos F, Razbash S, Wang E, Yasmeen F (2021).
forecast: Forecasting functions for time series and linear models. R
package version 8.14, <URL: https://pkg.robjhyndman.com/forecast/>.
Hyndman RJ, Khandakar Y (2008). 'Automatic time series forecasting: the forecast package for R.' Journal of Statistical Software, 26 (3), 1-22. <URL: https://www.jstatsoft.org/article/view/v027i03>.
Assimakopoulos, V. and Nikolopoulos, K. (2000). The theta model: a decomposition approach to forecasting. International Journal of Forecasting 16, 521-530.
Hyndman, R.J., and Billah, B. (2003) Unmasking the Theta method. International J. Forecasting, 19, 287-290.
require(forecast) ## Not run: print(ahead::eatf(WWWusage, method = "EAT", weights = c(0.5, 0, 0.5))) print(ahead::eatf(WWWusage, method = "EAT")) obj <- ahead::eatf(WWWusage, method = "EAT", weights = c(0, 0.5, 0.5), h=10, type_pi = "T") plot(obj) obj <- ahead::eatf(WWWusage, method = "EAT", weights = c(0, 0.5, 0.5), h=10, type_pi="A") plot(obj) ## End(Not run) par(mfrow=c(3, 2)) plot(ahead::eatf(USAccDeaths, h=10, level=95)) plot(ahead::eatf(AirPassengers, h=10, level=95, type_pi = "T")) plot(ahead::eatf(lynx, h=10, level=95, type_pi = "A")) plot(ahead::eatf(WWWusage, h=10, level=95, type_pi = "E")) plot(ahead::eatf(Nile, h=10, level=95)) plot(ahead::eatf(fdeaths, h=10, level=95))require(forecast) ## Not run: print(ahead::eatf(WWWusage, method = "EAT", weights = c(0.5, 0, 0.5))) print(ahead::eatf(WWWusage, method = "EAT")) obj <- ahead::eatf(WWWusage, method = "EAT", weights = c(0, 0.5, 0.5), h=10, type_pi = "T") plot(obj) obj <- ahead::eatf(WWWusage, method = "EAT", weights = c(0, 0.5, 0.5), h=10, type_pi="A") plot(obj) ## End(Not run) par(mfrow=c(3, 2)) plot(ahead::eatf(USAccDeaths, h=10, level=95)) plot(ahead::eatf(AirPassengers, h=10, level=95, type_pi = "T")) plot(ahead::eatf(lynx, h=10, level=95, type_pi = "A")) plot(ahead::eatf(WWWusage, h=10, level=95, type_pi = "E")) plot(ahead::eatf(Nile, h=10, level=95)) plot(ahead::eatf(fdeaths, h=10, level=95))
dynrmf)Fit univariate time series using caret ML model (for use with dynrmf)
fit_func( x, y, method = "ranger", initial_window = 10L, horizon = 10L, fixed_window = FALSE, tune_length = 5, summary_function = NULL, verbose = TRUE )fit_func( x, y, method = "ranger", initial_window = 10L, horizon = 10L, fixed_window = FALSE, tune_length = 5, summary_function = NULL, verbose = TRUE )
x |
A matrix of predictors |
y |
A vector of responses |
method |
The caret method to use for fitting the model |
initial_window |
The initial window size |
horizon |
The forecast horizon |
fixed_window |
Whether to use a fixed window size |
tune_length |
Length of the tuning grid |
verbose |
Whether to print the model summary |
A model object
Fit and forecast for benchmarking purposes
fitforecast( y, h = NULL, pct_train = 0.9, pct_calibration = 0.5, method = c("thetaf", "arima", "ets", "te", "tbats", "tslm", "dynrmf", "ridge2f", "naive", "snaive"), level = 95, B = 1000L, seed = 17223L, graph = TRUE, conformalize = FALSE, type_calibration = c("splitconformal", "cv1", "loocv"), gap = 3L, agg = c("mean", "median"), vol = c("constant", "garch"), type_sim = c("kde", "surrogate", "bootstrap"), ... )fitforecast( y, h = NULL, pct_train = 0.9, pct_calibration = 0.5, method = c("thetaf", "arima", "ets", "te", "tbats", "tslm", "dynrmf", "ridge2f", "naive", "snaive"), level = 95, B = 1000L, seed = 17223L, graph = TRUE, conformalize = FALSE, type_calibration = c("splitconformal", "cv1", "loocv"), gap = 3L, agg = c("mean", "median"), vol = c("constant", "garch"), type_sim = c("kde", "surrogate", "bootstrap"), ... )
y |
A univariate time series of class |
h |
Forecasting horizon (default is |
pct_train |
Percentage of data in the training set, when |
pct_calibration |
Percentage of data in the calibration set for conformal prediction |
method |
For now "thetaf" (default), "arima", "ets", "tbats", "tslm", "dynrmf" (from ahead), "ridge2f" (from ahead), "naive", "snaive" |
level |
Confidence level for prediction intervals in %, default is 95 |
B |
Number of bootstrap replications or number of simulations (yes, 'B' is unfortunate) |
seed |
Reproducibility seed |
graph |
Plot or not? |
conformalize |
Calibrate or not? |
type_calibration |
"splitconformal" (default conformal method), "cv1" (do not use), "loocv" (do not use) |
gap |
length of training set for loocv conformal (do not use) |
agg |
"mean" or "median" (aggregation method) for |
vol |
"constant" or "garch" (type of volatility modeling for calibrated residuals) |
type_sim |
"kde", "surrogate", "bootstrap" (type of simulation for calibrated residuals) |
... |
additional parameters |
an object of class 'forecast' with additional information
par(mfrow=c(2, 2)) obj1 <- ahead::fitforecast(AirPassengers) obj2 <- ahead::fitforecast(AirPassengers, conformalize = TRUE) plot(AirPassengers) plot(obj1) obj2$plot() obj2$plot("simulations")par(mfrow=c(2, 2)) obj1 <- ahead::fitforecast(AirPassengers) obj2 <- ahead::fitforecast(AirPassengers, conformalize = TRUE) plot(AirPassengers) plot(obj1) obj2$plot() obj2$plot("simulations")
This function allows to call any function "of class forecast" in a unified way.
genericforecast(FUN, y, h, level = 95, ...)genericforecast(FUN, y, h, level = 95, ...)
FUN |
A forecasting function. |
y |
A time series ( |
h |
Forecasting horizon. |
level |
The confidence level. |
... |
Additional arguments to be passed to the forecasting function. |
An object of class forecast.
y <- fdeaths h <- 25L plot(genericforecast(FUN=forecast::thetaf, y, h)) plot(genericforecast(FUN=ahead::dynrmf, y, h)) plot(genericforecast(FUN=forecast::tbats, y=y, h=h, use.box.cox = TRUE, use.trend=FALSE))y <- fdeaths h <- 25L plot(genericforecast(FUN=forecast::thetaf, y, h)) plot(genericforecast(FUN=ahead::dynrmf, y, h)) plot(genericforecast(FUN=forecast::tbats, y=y, h=h, use.box.cox = TRUE, use.trend=FALSE))
Calculate returns or log-returns for multivariate time series
getreturns(x, type = c("basic", "log"))getreturns(x, type = c("basic", "log"))
x |
Multivariate time series |
type |
Type of return: basic return ("basic") or log-return ("log") |
The returns
returns <- getreturns(EuStockMarkets) log_returns <- getreturns(EuStockMarkets, type = "log") par(mfrow=c(1, 3)) matplot(EuStockMarkets, type = 'l', main = "Closing Prices of \n European stocks (1991-1998)", xlab = "time") matplot(returns, type = 'l', main = "Returns", xlab = "time") matplot(log_returns, type = 'l', main = "Log-returns", xlab = "time")returns <- getreturns(EuStockMarkets) log_returns <- getreturns(EuStockMarkets, type = "log") par(mfrow=c(1, 3)) matplot(EuStockMarkets, type = 'l', main = "Closing Prices of \n European stocks (1991-1998)", xlab = "time") matplot(returns, type = 'l', main = "Returns", xlab = "time") matplot(log_returns, type = 'l', main = "Log-returns", xlab = "time")
Obtain simulations (when relevant) from a selected time series
getsimulations(obj, selected_series, transpose = FALSE)getsimulations(obj, selected_series, transpose = FALSE)
obj |
result from ridge2f (multivariate time series forecast with simulations) |
selected_series |
name of the time series selected |
transpose |
return a transposed time series |
require(fpp) obj <- ahead::ridge2f(fpp::insurance, h = 7, type_pi = "bootstrap", B = 5) print(getsimulations(obj, selected_series = "TV.advert")) print(getsimulations(obj, selected_series = "Quotes")) print(getsimulations(obj, selected_series = "TV.advert", transpose = TRUE)) print(getsimulations(obj, selected_series = "Quotes", transpose = TRUE))require(fpp) obj <- ahead::ridge2f(fpp::insurance, h = 7, type_pi = "bootstrap", B = 5) print(getsimulations(obj, selected_series = "TV.advert")) print(getsimulations(obj, selected_series = "Quotes")) print(getsimulations(obj, selected_series = "TV.advert", transpose = TRUE)) print(getsimulations(obj, selected_series = "Quotes", transpose = TRUE))
Loess forecasting
loessf( y, h = 5, level = 95, span = 0.75, degree = 2, type_pi = c("bootstrap", "blockbootstrap", "movingblockbootstrap"), b = NULL, B = 250, type_aggregation = c("mean", "median"), seed = 123 )loessf( y, h = 5, level = 95, span = 0.75, degree = 2, type_pi = c("bootstrap", "blockbootstrap", "movingblockbootstrap"), b = NULL, B = 250, type_aggregation = c("mean", "median"), seed = 123 )
y |
A numeric vector or time series of class |
h |
Forecasting horizon |
level |
Confidence level for prediction intervals |
span |
the parameter which controls the degree of smoothing |
degree |
the degree of the polynomials to be used, normally 1 or 2. (Degree 0 is also allowed, but see |
type_pi |
Type of prediction interval currently (independent) "bootstrap", (circular) "blockbootstrap", or "movingblockbootstrap" |
b |
block length for circular block bootstrap |
B |
number of bootstrap replications |
type_aggregation |
Type of aggregation, ONLY for bootstrapping; either "mean" or "median" |
seed |
reproducibility seed |
An object of class "forecast"; a list containing the following elements:
model |
A list containing information about the fitted model |
method |
The name of the forecasting method as a character string |
mean |
Point forecasts for the time series |
lower |
Lower bound for prediction interval |
upper |
Upper bound for prediction interval |
x |
The original time series |
residuals |
Residuals from the fitted model |
sims |
Model simulations for bootstrapping |
par(mfrow = c(3, 1)) plot(loessf(Nile, h=20, level=95, B=10)) plot(loessf(Nile, h=20, level=95, B=10, type_pi = "blockbootstrap")) plot(loessf(Nile, h=20, level=95, B=10, type_pi = "movingblockbootstrap"))par(mfrow = c(3, 1)) plot(loessf(Nile, h=20, level=95, B=10)) plot(loessf(Nile, h=20, level=95, B=10, type_pi = "blockbootstrap")) plot(loessf(Nile, h=20, level=95, B=10, type_pi = "movingblockbootstrap"))
LOOCV for Random Vector functional link network model with 2 regularization parameters
loocvridge2f( y, xreg = NULL, h = 5, level = 95, lags = 1, nb_hidden = 5, nodes_sim = c("sobol", "halton", "unif"), activ = c("relu", "sigmoid", "tanh", "leakyrelu", "elu", "linear"), a = 0.01, lambda_1 = 0.1, lambda_2 = 0.1, dropout = 0, type_forecast = c("recursive", "direct"), type_pi = c("gaussian", "bootstrap", "blockbootstrap", "movingblockbootstrap", "rvinecopula", "splitconformal"), block_length = NULL, margins = c("gaussian", "empirical", "student"), seed = 1, B = 100L, type_aggregation = c("mean", "median"), centers = NULL, type_clustering = c("kmeans", "hclust"), ym = NULL, cl = 1L, show_progress = TRUE, ... )loocvridge2f( y, xreg = NULL, h = 5, level = 95, lags = 1, nb_hidden = 5, nodes_sim = c("sobol", "halton", "unif"), activ = c("relu", "sigmoid", "tanh", "leakyrelu", "elu", "linear"), a = 0.01, lambda_1 = 0.1, lambda_2 = 0.1, dropout = 0, type_forecast = c("recursive", "direct"), type_pi = c("gaussian", "bootstrap", "blockbootstrap", "movingblockbootstrap", "rvinecopula", "splitconformal"), block_length = NULL, margins = c("gaussian", "empirical", "student"), seed = 1, B = 100L, type_aggregation = c("mean", "median"), centers = NULL, type_clustering = c("kmeans", "hclust"), ym = NULL, cl = 1L, show_progress = TRUE, ... )
y |
A multivariate time series of class |
xreg |
External regressors. A data.frame (preferred) or a |
h |
Forecasting horizon |
level |
Confidence level for prediction intervals |
lags |
Number of lags |
|
Number of nodes in hidden layer |
|
nodes_sim |
Type of simulation for nodes in the hidden layer |
activ |
Activation function |
a |
Hyperparameter for activation function "leakyrelu", "elu" |
lambda_1 |
Regularization parameter for original predictors |
lambda_2 |
Regularization parameter for transformed predictors |
dropout |
dropout regularization parameter (dropping nodes in hidden layer) |
type_forecast |
Recursive or direct forecast |
type_pi |
Type of prediction interval currently "gaussian", "bootstrap", "blockbootstrap", "movingblockbootstrap", "splitconformal" (very experimental right now), "rvinecopula" (with Gaussian margins for now, Student-t coming soon) |
block_length |
Length of block for circular or moving block bootstrap |
margins |
Distribution of margins: "gaussian", "empirical", "student" (postponed or
never) for |
seed |
Reproducibility seed for random stuff |
B |
Number of bootstrap replications or number of simulations (yes, 'B' is unfortunate) |
type_aggregation |
Type of aggregation, ONLY for bootstrapping; either "mean" or "median" |
centers |
Number of clusters for |
type_clustering |
"kmeans" (K-Means clustering) or "hclust" (Hierarchical clustering) |
ym |
Univariate time series ( |
cl |
An integer; the number of clusters for parallel execution, for bootstrap |
show_progress |
A boolean; show progress bar for bootstrapping? Default is TRUE. |
... |
An object of class "mtsforecast"; a list containing the following elements:
method |
The name of the forecasting method as a character string |
mean |
Point forecasts for the time series |
lower |
Lower bound for prediction interval |
upper |
Upper bound for prediction interval |
sims |
Model simulations for bootstrapping (basic, or block) |
x |
The original time series |
residuals |
Residuals from the fitted model |
coefficients |
Regression coefficients for |
T. Moudiki
Moudiki, T., Planchet, F., & Cousin, A. (2018).
Multiple time series forecasting using quasi-randomized
functional link neural networks. Risks, 6(1), 22.
require(fpp) print(ahead::loocvridge2f(fpp::insurance)) print(ahead::loocvridge2f(fpp::usconsumption)) #foo <- function(xx) ahead::loocvridge2f(fpp::insurance, lambda_1=10^xx[1], lambda_2=10^xx[2]) #(opt <- stats::nlminb(objective=foo, lower=c(-10,-10), upper=c(10,10), start=c(0, 0))) #print(ahead::loocvridge2f(fpp::insurance, lambda_1=10^opt$par[1], lambda_2=10^opt$par[2]))require(fpp) print(ahead::loocvridge2f(fpp::insurance)) print(ahead::loocvridge2f(fpp::usconsumption)) #foo <- function(xx) ahead::loocvridge2f(fpp::insurance, lambda_1=10^xx[1], lambda_2=10^xx[2]) #(opt <- stats::nlminb(objective=foo, lower=c(-10,-10), upper=c(10,10), start=c(0, 0))) #print(ahead::loocvridge2f(fpp::insurance, lambda_1=10^opt$par[1], lambda_2=10^opt$par[2]))
Produces plots for the results of a forecast combination method. Either
an actual vs. fitted plot (which = 1) or a barplot of the combination weights
(which = 2).
## S3 method for class 'foreccomb_res' plot(x, which = 1, ...)## S3 method for class 'foreccomb_res' plot(x, which = 1, ...)
x |
An object of class 'foreccomb_res'. |
which |
Type of plot: 1 = actual vs. fitted, 2 = combination weights. |
... |
Other arguments passing to |
A plot for the foreccomb_res class.
adapted from Christoph E. Weiss and Gernot R. Roetzer (ForecastComb)
foreccomb,
summary.foreccomb_res
Plot multivariate time series forecast or residuals
## S3 method for class 'mtsforecast' plot(x, selected_series, type = c("pi", "dist", "sims"), level = 95, ...)## S3 method for class 'mtsforecast' plot(x, selected_series, type = c("pi", "dist", "sims"), level = 95, ...)
x |
result from |
selected_series |
name of the time series selected for plotting |
type |
"pi": basic prediction intervals; "dist": a distribution of predictions; "sims": the simulations |
level |
confidence levels for prediction intervals |
... |
additional parameters to be passed to |
require(fpp) fit_obj_VAR <- ahead::varf(fpp::insurance, lags = 2, h = 10, level = 95) fit_obj_ridge2 <- ahead::ridge2f(fpp::insurance, lags = 2, h = 10, level = 95) par(mfrow=c(2, 2)) plot(fit_obj_VAR, "Quotes") plot(fit_obj_VAR, "TV.advert") plot(fit_obj_ridge2, "Quotes") plot(fit_obj_ridge2, "TV.advert") obj <- ahead::ridge2f(fpp::insurance, h = 10, type_pi = "blockbootstrap", block_length=5, B = 10) par(mfrow=c(1, 2)) plot(obj, selected_series = "Quotes", type = "sims", main = "Predictive simulation for Quotes") plot(obj, selected_series = "TV.advert", type = "sims", main = "Predictive simulation for TV.advert") par(mfrow=c(1, 2)) plot(obj, selected_series = "Quotes", type = "dist", main = "Predictive simulation for Quotes") plot(obj, selected_series = "TV.advert", type = "dist", main = "Predictive simulation for TV.advert")require(fpp) fit_obj_VAR <- ahead::varf(fpp::insurance, lags = 2, h = 10, level = 95) fit_obj_ridge2 <- ahead::ridge2f(fpp::insurance, lags = 2, h = 10, level = 95) par(mfrow=c(2, 2)) plot(fit_obj_VAR, "Quotes") plot(fit_obj_VAR, "TV.advert") plot(fit_obj_ridge2, "Quotes") plot(fit_obj_ridge2, "TV.advert") obj <- ahead::ridge2f(fpp::insurance, h = 10, type_pi = "blockbootstrap", block_length=5, B = 10) par(mfrow=c(1, 2)) plot(obj, selected_series = "Quotes", type = "sims", main = "Predictive simulation for Quotes") plot(obj, selected_series = "TV.advert", type = "sims", main = "Predictive simulation for TV.advert") par(mfrow=c(1, 2)) plot(obj, selected_series = "Quotes", type = "dist", main = "Predictive simulation for Quotes") plot(obj, selected_series = "TV.advert", type = "dist", main = "Predictive simulation for TV.advert")
dynrmf)Predict univariate time series using caret ML model(for use with dynrmf)
predict_func(obj, newx)predict_func(obj, newx)
x |
A matrix of predictors |
y |
A vector of responses |
method |
The caret method to use for fitting the model |
verbose |
Whether to print the model summary |
A model object
prediction method for class ‘foreccomb_res’. Uses the previously created forecast combination
result to predict the combination for a newly provided prediction dataset.
## S3 method for class 'foreccomb_res' predict(object, newpreds, newobs = NULL, simplify = TRUE, byrow = FALSE, ...)## S3 method for class 'foreccomb_res' predict(object, newpreds, newobs = NULL, simplify = TRUE, byrow = FALSE, ...)
object |
An object of class 'foreccomb'. Contains training set (actual values + matrix of model forecasts) and optionally a test set. |
newpreds |
A matrix or multivariate time series; contains individual model forecasts if a test set is used (optional). Does not
require specification of |
newobs |
A vector or univariate time series; contains ‘actual values’ if a test set is used (optional). |
simplify |
logical. The default ( |
byrow |
logical. The default ( |
... |
potential further arguments (require by generic) |
Adapted from Christoph E. Weiss and Gernot R. Roetzer (ForecastComb)
Random Vector functional link network model with 2 regularization parameters
ridge2f( y, h = 5, level = 95, xreg = NULL, lags = 1, nb_hidden = 5, nodes_sim = c("sobol", "halton", "unif"), activ = c("relu", "sigmoid", "tanh", "leakyrelu", "elu", "linear"), a = 0.01, lambda_1 = 0.1, lambda_2 = 0.1, dropout = 0, type_forecast = c("recursive", "direct"), type_pi = c("gaussian", "bootstrap", "blockbootstrap", "movingblockbootstrap", "rvinecopula"), block_length = NULL, margins = c("gaussian", "empirical"), seed = 1, B = 100L, type_aggregation = c("mean", "median"), centers = NULL, type_clustering = c("kmeans", "hclust"), ym = NULL, cl = 1L, show_progress = TRUE, ... )ridge2f( y, h = 5, level = 95, xreg = NULL, lags = 1, nb_hidden = 5, nodes_sim = c("sobol", "halton", "unif"), activ = c("relu", "sigmoid", "tanh", "leakyrelu", "elu", "linear"), a = 0.01, lambda_1 = 0.1, lambda_2 = 0.1, dropout = 0, type_forecast = c("recursive", "direct"), type_pi = c("gaussian", "bootstrap", "blockbootstrap", "movingblockbootstrap", "rvinecopula"), block_length = NULL, margins = c("gaussian", "empirical"), seed = 1, B = 100L, type_aggregation = c("mean", "median"), centers = NULL, type_clustering = c("kmeans", "hclust"), ym = NULL, cl = 1L, show_progress = TRUE, ... )
y |
A univariate of multivariate time series of class |
h |
Forecasting horizon |
level |
Confidence level for prediction intervals |
xreg |
External regressors. A data.frame (preferred) or a |
lags |
Number of lags |
|
Number of nodes in hidden layer |
|
nodes_sim |
Type of simulation for nodes in the hidden layer |
activ |
Activation function |
a |
Hyperparameter for activation function "leakyrelu", "elu" |
lambda_1 |
Regularization parameter for original predictors |
lambda_2 |
Regularization parameter for transformed predictors |
dropout |
dropout regularization parameter (dropping nodes in hidden layer) |
type_forecast |
Recursive or direct forecast |
type_pi |
Type of prediction interval currently "gaussian", "bootstrap", "blockbootstrap", "movingblockbootstrap", "rvinecopula" (with Gaussian and empirical margins for now) |
block_length |
Length of block for circular or moving block bootstrap |
margins |
Distribution of margins: "gaussian", "empirical" for |
seed |
Reproducibility seed for random stuff |
B |
Number of bootstrap replications or number of simulations (yes, 'B' is unfortunate) |
type_aggregation |
Type of aggregation, ONLY for bootstrapping; either "mean" or "median" |
centers |
Number of clusters for |
type_clustering |
"kmeans" (K-Means clustering) or "hclust" (Hierarchical clustering) |
ym |
Univariate time series ( |
cl |
An integer; the number of clusters for parallel execution, for bootstrap |
show_progress |
A boolean; show progress bar for bootstrapping? Default is TRUE. |
... |
An object of class "mtsforecast"; a list containing the following elements:
method |
The name of the forecasting method as a character string |
mean |
Point forecasts for the time series |
lower |
Lower bound for prediction interval |
upper |
Upper bound for prediction interval |
sims |
Model simulations for bootstrapping (basic, or block) |
x |
The original time series |
residuals |
Residuals from the fitted model |
coefficients |
Regression coefficients for |
T. Moudiki
Moudiki, T., Planchet, F., & Cousin, A. (2018).
Multiple time series forecasting using quasi-randomized
functional link neural networks. Risks, 6(1), 22.
require(fpp) print(ahead::ridge2f(fpp::insurance)$mean) print(ahead::ridge2f(fpp::usconsumption)$lower) res <- ahead::ridge2f(fpp::insurance, h=10, lags=2) par(mfrow=c(1, 2)) plot(res, "Quotes") plot(res, "TV.advert") # include a trend (just for the example) xreg <- as.numeric(time(fpp::insurance)) res2 <- ahead::ridge2f(fpp::insurance, xreg=xreg, h=10, lags=2) par(mfrow=c(1, 2)) plot(res2, "Quotes") plot(res2, "TV.advert") # block bootstrap xreg <- as.numeric(time(fpp::insurance)) res3 <- ahead::ridge2f(fpp::insurance, xreg=xreg, h=10, lags=1L, type_pi = "bootstrap", B=10) res5 <- ahead::ridge2f(fpp::insurance, xreg=xreg, h=10, lags=1L, type_pi = "blockbootstrap", B=10, block_length = 4) print(res3$sims[[2]]) print(res5$sims[[2]]) par(mfrow=c(2, 2)) plot(res3, "Quotes") plot(res3, "TV.advert") plot(res5, "Quotes") plot(res5, "TV.advert") res4 <- ahead::ridge2f(fpp::usconsumption, h=20, lags=2L, lambda_2=1) par(mfrow=c(1, 2)) plot(res4, "income") plot(res4, "consumption") # moving block bootstrap xreg <- as.numeric(time(fpp::insurance)) res6 <- ahead::ridge2f(fpp::insurance, xreg=xreg, h=10, lags=1L, type_pi = "movingblockbootstrap", B=10, block_length = 4) print(res6$sims[[2]]) par(mfrow=c(1, 2)) plot(res6, "Quotes") plot(res6, "TV.advert")require(fpp) print(ahead::ridge2f(fpp::insurance)$mean) print(ahead::ridge2f(fpp::usconsumption)$lower) res <- ahead::ridge2f(fpp::insurance, h=10, lags=2) par(mfrow=c(1, 2)) plot(res, "Quotes") plot(res, "TV.advert") # include a trend (just for the example) xreg <- as.numeric(time(fpp::insurance)) res2 <- ahead::ridge2f(fpp::insurance, xreg=xreg, h=10, lags=2) par(mfrow=c(1, 2)) plot(res2, "Quotes") plot(res2, "TV.advert") # block bootstrap xreg <- as.numeric(time(fpp::insurance)) res3 <- ahead::ridge2f(fpp::insurance, xreg=xreg, h=10, lags=1L, type_pi = "bootstrap", B=10) res5 <- ahead::ridge2f(fpp::insurance, xreg=xreg, h=10, lags=1L, type_pi = "blockbootstrap", B=10, block_length = 4) print(res3$sims[[2]]) print(res5$sims[[2]]) par(mfrow=c(2, 2)) plot(res3, "Quotes") plot(res3, "TV.advert") plot(res5, "Quotes") plot(res5, "TV.advert") res4 <- ahead::ridge2f(fpp::usconsumption, h=20, lags=2L, lambda_2=1) par(mfrow=c(1, 2)) plot(res4, "income") plot(res4, "consumption") # moving block bootstrap xreg <- as.numeric(time(fpp::insurance)) res6 <- ahead::ridge2f(fpp::insurance, xreg=xreg, h=10, lags=1L, type_pi = "movingblockbootstrap", B=10, block_length = 4) print(res6$sims[[2]]) par(mfrow=c(1, 2)) plot(res6, "Quotes") plot(res6, "TV.advert")
summary method for class ‘foreccomb_res’. Includes information about combination method,
combination weights assigned to the individual forecast models, as well as an accuracy evaluation of the combined
forecast.
## S3 method for class 'foreccomb_res' summary(object, ...) ## S3 method for class 'foreccomb_res_summary' print(x, ...)## S3 method for class 'foreccomb_res' summary(object, ...) ## S3 method for class 'foreccomb_res_summary' print(x, ...)
object |
An object of class 'foreccomb'. Contains training set (actual values + matrix of model forecasts) and optionally a test set. |
... |
potential further arguments (require by generic) |
x |
An object of class 'foreccomb'. Contains training set (actual values + matrix of model forecasts) and optionally a test set. |
Christoph E. Weiss and Gernot R. Roetzer
foreccomb,
plot.foreccomb_res,
Vector Autoregressive model adapted from vars::VAR (only for benchmarking)
varf( y, h = 5, level = 95, lags = 1, type_VAR = c("const", "trend", "both", "none"), ... )varf( y, h = 5, level = 95, lags = 1, type_VAR = c("const", "trend", "both", "none"), ... )
y |
A multivariate time series of class |
h |
Forecasting horizon |
level |
Confidence level for prediction intervals |
lags |
Number of lags |
type_VAR |
Type of deterministic regressors to include. |
... |
Additional parameters to be passed to vars::VAR. |
An object of class "mtsforecast"; a list containing the following elements:
method |
The name of the forecasting method as a character string |
mean |
Point forecasts for the time series |
lower |
Lower bound for prediction interval |
upper |
Upper bound for prediction interval |
x |
The original time series |
residuals |
Residuals from the fitted model |
T. Moudiki
Bernhard Pfaff (2008). VAR, SVAR and SVEC Models: Implementation
Within R Package vars. Journal of Statistical Software 27(4). URL
http://www.jstatsoft.org/v27/i04/.
Pfaff, B. (2008) Analysis of Integrated and Cointegrated Time Series with R. Second Edition. Springer, New York. ISBN 0-387-27960-1
require(fpp) print(varf(fpp::insurance, lags=2, h=10)) res <- varf(fpp::usconsumption, h=20, lags=2) par(mfrow=c(1, 2)) plot(res, "consumption") plot(res, "income")require(fpp) print(varf(fpp::insurance, lags=2, h=10)) res <- varf(fpp::usconsumption, h=20, lags=2) par(mfrow=c(1, 2)) plot(res, "consumption") plot(res, "income")