Type: Package
Title: Machine Learning Time Series Forecasting
Version: 1.1.2
Date: 2025-06-28
Author: Ho Tsung-wu [aut, cre]
Maintainer: Ho Tsung-wu <tsungwu@ntnu.edu.tw>
Description: Compute onestep and multistep time series forecasts for machine learning models.
License: GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]
LazyData: TRUE
LazyLoad: yes
Depends: R (≥ 3.5)
Imports: caret,zoo
Suggests: forecast, h2o, kernlab, lubridate, timeSeries, timeDate, xts
NeedsCompilation: no
Packaged: 2025-06-28 11:10:51 UTC; badal
Repository: CRAN
Date/Publication: 2025-06-28 11:40:02 UTC

Accuracy measures for a forecast model

Description

Returns range of summary measures of the forecast accuracy. Except MAAPE, all measures are defined and discussed in Hyndman and Koehler (2006).

Usage

Accuracy(f,x)

Arguments

f

A time series forecasting object generated by iForecast.

x

Actual values of the same length as the time series object of f.

Details

The measures calculated are:

Except MAAPE, by default, see Hyndman and Koehler (2006) and Hyndman and Athanasopoulos (2014, Section 2.5) for further details. For MAAPE, please see Kim and Kim (2016).

Value

Matrix giving forecast accuracy measures.

Author(s)

Ho Tsung-wu <tsungwu@ntnu.edu.tw>, College of Management, National Taiwan Normal University.

References

Hyndman, R.J. and Koehler, A.B. (2006) "Another look at measures of forecast accuracy". International Journal of Forecasting, 22(4), 679-688.
Hyndman, R.J. and Athanasopoulos, G. (2018) "Forecasting: principles and practice", 2nd ed., OTexts, Melbourne, Australia. Section 3.4 "Evaluating forecast accuracy".<https://otexts.com/fpp2/accuracy.html>
Kim Sungil and Heeyoung Kim (2016) "A new metric of absolute percentage error for intermittent demand forecasts", International Journal of Forecasting,32(3),669-679. <https://doi.org/10.1016/j.ijforecast.2015.12.003>.

Examples


tmp0=timeSeries::as.timeSeries(ts(rnorm(800),start=c(1960,1),freq=12))
fit1 <- timeSeries::as.timeSeries(forecast::rwf(tmp0[1:700,1],h=100)$mean)
Accuracy(f=fit1,x=tmp0[701:800,1])

Economic and Financial Data Sets

Description

ES_15m is 15-min realized absolute variance of e-mini S&P 500. macrodata contains monthly US unemployment(unrate), ES_Daily is daily realized absolute variance of e-mini S&P 500. macrodata contains monthly US unemployment(unrate) and and year-to-year changes in three regional business cycle indices (OECD, NAFTA, and G7). bc contains monthly business cycle data, bc is binary indicator(1=recession, 2=boom) of Taiwan's business cycle phases, IPI_TWN is industrial production index of Taiwan, LD_OECD, LD_G7, and LD_NAFTA are leading indicators of OECD, G7 and NAFTA regions; all four are monthly rate of changes.

Usage

data(ES_15m)
data(macrodata)
data(ES_Daily)
data(bc)

Value

an object of class "zoo".

Extract predictions and class probabilities from train objects

Description

It generates both the static and recursive time series plots of machine learning prediction object generated by ttsCaret, ttsAutoML and ttsLSTM.

Usage

iForecast(Model,newdata,Type,n.ahead)

Arguments

Model

Object of trained model.

newdata

The dataset for pediction, the column names must be the same as the trained data.Not required if type="dynamic".

Type

If Type="static", it computes the (static) forecasting values of insample model fit. If Type="dynamic", it recursively computes the forecasting values ahead.

n.ahead

For Type="dynamic", it is the number of forecasting periods ahead. Not required if type="static".

Details

This function generates forecasts of tts.caret,and tts.autoML.

Value

prediction

The forecasted time series target variable. For binary case, it returns both porbabilities and class.

Author(s)

Ho Tsung-wu <tsungwu@ntnu.edu.tw>, College of Management, National Taiwan Normal University.

Examples

# Cross-validation takes time, example below is commented.
## Machine Learning by library(caret)
#Case 1. Low frequency, regression type
data("macrodata")
dep <- macrodata[569:669,"unrate",drop=FALSE]
ind <- macrodata[569:669,-1,drop=FALSE]
train.end <- "2018-12-01"# Choosing the end dating of train

models <- c("svm","rf","rpart","gbm","nb")[1]
type <- c("none","trend","season","both")[1]
# output <- tts.caret(y=dep, x=ind, arOrder=c(1), xregOrder=c(1),
# method="smv", tuneLength =1,
# train.end, type=type,resampling="cv",preProcess = "center")

# testData1 <- window(output$dataused,start="2019-01-01",end=end(output$dataused))
#P1=iForecast(Model=output,Type="static",newdata=testData1)
#P2=iForecast(Model=output,Type="dynamic",n.ahead=7)

#tail(cbind(testData1[,1],P1))
#tail(cbind(testData1[,1],P2))

#Case 2. Low frequency, binary type
data(bc) #binary dependent variable, business cycle phases
dep=bc[,1,drop=FALSE]
ind=bc[,-1]

train.end=as.character(rownames(dep))[as.integer(nrow(dep)*0.8)]
test.start=as.character(rownames(dep))[as.integer(nrow(dep)*0.8)+1]

#output = tts.caret(y=dep, x=ind, arOrder=c(1), xregOrder=c(1), method="nb",
#                   tuneLength =10, train.end, type=type)

#testData1=window(output$dataused,start=test.start,end=end(output$dataused))

#head(output$dataused)
#P1=iForecast(Model=output,Type="static",newdata=testData1)
#P2=iForecast(Model=output,Type="dynamic",n.ahead=7)

#tail(cbind(testData1[,1],P1),10)
#tail(cbind(testData1[,1],P2),10)

Defunct functions in package ‘iForecast’

Description

These functions are defunct and no longer available.

Details

Defunct function is: ttsAutoML New function is: tts.autoML

Defunct functions in package ‘iForecast’

Description

These functions are defunct and no longer available.

Details

Defunct function is: ttsCaret New function is: tts.caret

Defunct functions in package ‘iForecast’

Description

These functions are defunct and no longer available.

Details

Defunct functions are: ttsLSTM

Produce multistep forecasts from machine learning VAR

Description

It generates multistep forecasts of machine learning VAR.

Usage


iForecast.var(object, n.ahead)

Arguments

object

The object generated by tts.var.

n.ahead

The number of out-of-sample forecasting periods. If n.ahead=1, it is one-step forecast; if n.ahead>1, it computes multistep forecasts by recursive method.

Details

This function generates multistep forecasts of machine learning VAR.

Author(s)

Ho Tsung-wu <tsungwu@ntnu.edu.tw>, College of Management, National Taiwan Normal University.

Examples

data(macrodata)
y=timeSeries::as.timeSeries(macrodata[,-1])
VLD=window(y,start="2019-01-01",end=end(y))
#OUT1=tts.var(data=y,
#             p=3,
#             method="enet",
#             train.end="2018-12-01",
#             type=c("none","trend","season","both")[1])

#fcst_ml=iForecast.var(OUT1, n.ahead=nrow(VLD))

Rolling timeframe for time series anaysis

Description

It extracts time stamp from a timeSeries object and separates the time into in-sample training and out-of-sample validation ranges.

Usage

rollingWindows(x,estimation="18m",by = "1m")

Arguments

x

The time series object with timeSeries, xts, or zoo format of "

estimation

The range of insample estimation period, the default is 18 months(18m), where the k-fold cross-section is performed. Quarter, week and day are also supported (see example).

by

The range of out-of-sample validation/testing period, the default is 6 months(6m).Quarter, week and day are also supported (see example).

Details

This function is similar to the backtesting framework in portfolio analysis. Rolling windows fixes the origin and the training sample grows over time, moving windows can be achieved by placing window() on dependent variable at each iteration.

Value

window

The time labels of from and to

.

Author(s)

Ho Tsung-wu <tsungwu@ntnu.edu.tw>, College of Management, National Taiwan Normal University.

Examples

data(macrodata)
y=macrodata[,1,drop=FALSE]
timeframe=rollingWindows(y,estimation="300m",by="6m")
#estimation="300m", because macrodata is monthly
FROM=timeframe$from
TO=timeframe$to

data(ES_Daily)
y=ES_Daily[,1,drop=FALSE]
timeframe=rollingWindows(y,estimation  ="60w",by="1w")
#60 weeks(300+ days) as estimation window and move by 1 week(5+ days).

FROM=timeframe$from
TO=timeframe$to

y=ES_Daily[,1,drop=FALSE]
timeframe=rollingWindows(y,estimation  ="250d",by="10d")
#250-day as estimation window and move by 10 days.


# simulated quarterly data
tmp0=ts(rnorm(800),start=c(1900,1),freq=4)
tmp1=timeSeries::as.timeSeries(tmp0)
tmp2=zoo::as.zoo(tmp0)
tmp3=xts::as.xts(tmp0)
timeframe=rollingWindows(x=tmp3,estimation ="100q",by="12q")
FROM=timeframe$from
TO=timeframe$to

It applies the h2o.deeplearning of h2o to time series data

Description

It applies the deep learning rountine, specifically the Multilayer Perceptron(MLP), of H2o.ai to time series data.

Usage

tts.DeepLearning(y,x=NULL,train.end,arOrder=2,xregOrder=0,type,initial=TRUE)

Arguments

y

The time series object of the target variable, for example, timeSeries,xts, or zoo. Numerically,y must be real numbers for regression or integers for classification. Date format must be "

x

The time series matrix of input variables, timestamp is the same as y, maybe null.

train.end

The end date of training data, must be specificed. The default dates of train.start and test.end are the start and the end of input data; and the test.start is the 1-period next of train.end.

arOrder

The autoregressive order of the target variable, which may be sequentially specifed like arOrder=1:5; or discontinuous lags like arOrder=c(1,3,5); zero is not allowed.

xregOrder

The distributed lag structure of the input variables, which may be sequentially specifed like xregOrder=1:5; or discontinuous lags like xregOrder=c(0,3,5); zero is allowed since contemporaneous correlation is allowed.

type

The time dummies variables. We have four selection:
'none'=no other variables,
'trend'=inclusion of time dummy,
'season'=inclusion of seasonal dummies,
'both'=inclusion of both trend and season. No default.

initial

Whether to initialize h2o.init(). Default to "TRUE" and, to avoid multiple initiations, users had better change it to FALSE while training via rolling windows. See example below.

Details

This function calls the h2o.deeplearning function from package h2o to execute multilayer percentron learning.

Value

output

Output object generated by h2o.automl function of h2o.

arOrder

The autoregressive order of the target variable used.

dataused

The data used by arOrder, xregOrder

data

The complete data structure

TD

Time dummies used, inherited from 'type' in tts.DeepLearning

train.end

The same as the argument in tts.caret

Author(s)

Ho Tsung-wu <tsungwu@ntnu.edu.tw>, College of Management, National Taiwan Normal University.

Examples

# Computation takes time, example below is commented.
data("macrodata")
dep<-macrodata[,"unrate",drop=FALSE]
ind<-macrodata[,-1,drop=FALSE]

# Choosing the dates of training and testing data
train.end<-"2008-12-01"

#Must execute the commands below
#h2o::h2o.init()        # Initialize h2o
#invisible(h2o::h2o.no_progress()) # Turn off progress bars

# out <- tts.DeepLearning(y=dep, x=ind, train.end,arOrder=c(2,4),
# xregOrder=c(0,1,3),type="both",initial=FALSE)

#testData2 <- window(out$dataused,start="2009-01-01",end=end(out$dataused))
#P1<-iForecast(Model=out,Type="static",newdata=testData2)
#P2<-iForecast(Model=out,Type="dynamic",n.ahead=nrow(testData2))

#tail(cbind(testData2[,1],P1))
#tail(cbind(testData2[,1],P2))

#h2o::h2o.shutdown(promp=FALSE) # Remember to shutdown h2o when all works are finished.

Train time series by automatic machine learning of h2o provided by H2o.ai

Description

It applies the h2o.autoML of H2O.ai to time series data.

Usage

tts.autoML(y,x=NULL,train.end,arOrder=2,xregOrder=0,type,max_models = 20,
                      sort_metric="AUTO",stopping_metric = "AUTO",initial=TRUE)

Arguments

y

The time series object of the target variable, for example, timeSeries,xts, or zoo. Numerically,y must be real numbers for regression or integers for classification. Date format must be "

x

The time series matrix of input variables, timestamp is the same as y, maybe null.

train.end

The end date of training data, must be specificed. The default dates of train.start and test.end are the start and the end of input data; and the test.start is the 1-period next of train.end.

arOrder

The autoregressive order of the target variable, which may be sequentially specifed like arOrder=1:5; or discontinuous lags like arOrder=c(1,3,5); zero is not allowed.

xregOrder

The distributed lag structure of the input variables, which may be sequentially specifed like xregOrder=1:5; or discontinuous lags like xregOrder=c(0,3,5); zero is allowed since contemporaneous correlation is allowed.

type

The time dummies variables. We have four selection:
'none'=no other variables,
'trend'=inclusion of time dummy,
'season'=inclusion of seasonal dummies,
'both'=inclusion of both trend and season. No default.

max_models

Number of AutoML base models, default to 20.

sort_metric

Specifies the metric used to sort the Leaderboard by at the end of an AutoML run. Defaults to "AUTO", where 'AUC' (area under the ROC curve) for binary classification, 'mean_per_class_error' for multinomial classification, and 'deviance' for regression. Available options include:'MSE','RMSE','MAE','RMSLE','AUCPR' (area under the Precision-Recall curve)

stopping_metric

Specify the metric to use for early stopping. Defaults to "AUTO",where 'logloss' for classification and 'deviance' for regression. Besides, options are: 'MSE','RMSE','MAE','RMSLE','AUC','AUCPR','lift_top_group'

initial

Whether to initialize h2o.init(). Default to "TRUE" and, to avoid multiple initiations, users had better change it to FALSE while training via rolling windows. See example below.

Details

This function calls the h2o.automl function from package h2o to execute automatic machine learning estimation.

Value

output

Output object generated by h2o.automl function of h2o.

modelsUsed

AutoML Leaderboard object, which is a table returns the argument of 'max_models'.

arOrder

The autoregressive order of the target variable used.

dataused

The data used by arOrder, xregOrder

data

The complete data structure

TD

Time dummies used, inherited from 'type' in tts.autoML

train.end

The same as the argument in tts.caret

Author(s)

Ho Tsung-wu <tsungwu@ntnu.edu.tw>, College of Management, National Taiwan Normal University.

Examples

# Computation takes time, example below is commented.
data("macrodata")
dep<-macrodata[,"unrate",drop=FALSE]
ind<-macrodata[,-1,drop=FALSE]

# Choosing the dates of training and testing data
train.end<-"2008-12-01"

#autoML of H2O.ai must execute the commands below
#h2o::h2o.init()        # Initialize h2o
#invisible(h2o::h2o.no_progress()) # Turn off progress bars

# autoML <- tts.autoML(y=dep, x=ind, train.end,arOrder=c(2,4),
# xregOrder=c(0,1,3),type="both",initial=FALSE)
# print(autoML$modelsUsed,n=22) #View the AutoML Leaderboard

#testData2 <- window(autoML$dataused,start="2009-01-01",end=end(autoML$dataused))
#P1<-iForecast(Model=autoML,Type="static",newdata=testData2)
#P2<-iForecast(Model=autoML,Type="dynamic",n.ahead=nrow(testData2))

#tail(cbind(testData2[,1],P1))
#tail(cbind(testData2[,1],P2))

#h2o::h2o.shutdown(promp=FALSE) # Remember to shutdown h2o when all works are finished.

Train time series by caret and produce two types of time series forecasts: static and dynamic

Description

It generates both the static and dynamic time series plots of machine learning prediction object generated by package caret.

Usage


tts.caret(
  y,
  x=NULL,
  method,
  train.end,
  arOrder=2,
  xregOrder=0,
  type,
  tuneLength =10,
  preProcess = NULL,
  resampling="boot",
  Number=NULL,
  Repeat=NULL)

Arguments

y

The time series object of the target variable, for example, timeSeries,xts, or zoo. y can be either binary or continuous. Date format must be "

x

The time series matrix of input variables, timestamp is the same as y, maybe null.

method

The train_model_list of caret. While using this, make sure that the method allows regression. Methods in c("svm","rf","rpart","gamboost","BstLm","bstSm","blackboost") are feasible.

train.end

The end date of training data, must be specificed.The default dates of train.start and test.end are the start and the end of input data; and the test.start is the 1-period next of train.end.

arOrder

The autoregressive order of the target variable, which may be sequentially specifed like arOrder=1:5; or discontinuous lags like arOrder=c(1,3,5); zero is not allowed.

xregOrder

The distributed lag structure of the input variables, which may be sequentially specifed like xregOrder=0:5; or discontinuous lags like xregOrder=c(0,3,5); zero is allowed since contemporaneous correlation is allowed.

type

The time dummies variables. We have four selection:
"none"=no other variables,
"trend"=inclusion of time dummy,
"season"=inclusion of seasonal dummies,
"both"=inclusion of both trend and season. No default.

tuneLength

The same as the length specified in train function of package caret.

preProcess

Whether to pre-process the data, current possibilities are "BoxCox", "YeoJohnson", "expoTrans", "center", "scale", "range", "knnImpute", "bagImpute", "medianImpute", "pca", "ica" and "spatialSign".The default is no pre-processing.

resampling

The method for resampling, as trainControl function list in package caret. The default is "boot" for bootstrapping with 25 replications. Current choices are c("cv","boot","repeatedcv","LOOCV") where "cv" is K-fold CV with a default K=10 or specified by the "Number" below, "LOOCV" denotes the leave-one-out CV

Number

The number of K for K-Fold CV, default (NULL) is 10; for "boot" option, the default number of replications is 25

Repeat

The number for the repeatition for "repeatedcv".

Details

This function calls the train function of package caret to execute estimation. When execution finished, we compute two types of time series forecasts: static and recursive.

Value

output

Output object generated by train function of caret.

arOrder

The autoregressive order of the target variable used.

training.Pred

All tuned prediction values of training data, using besTunes to extract the best prediction.

dataused

The data used by arOrder, xregOrder, and type.

data

The complete data structure

TD

Time dummies used, inherited from type

train.end

The same as argument in tts.caret

Author(s)

Ho Tsung-wu <tsungwu@ntnu.edu.tw>, College of Management, National Taiwan Normal University.

Examples

# Cross-validation takes time, example below is commented.
## Machine Learning by library(caret)
library(zoo)
#Case 1. Low frequency
data("macrodata")
dep <- macrodata[569:669,"unrate",drop=FALSE]
ind <- macrodata[569:669,-1,drop=FALSE]
train.end <- "2018-12-01"# Choosing the end dating of train

models <- c("glm","knn","nnet","rpart","rf","svm","enet","gbm","lasso","bridge","nb")[2]
type <- c("none","trend","season","both")[1]
output <- tts.caret(y=dep,x=NULL, arOrder=c(1), xregOrder=c(1),
 method=models, tuneLength =1, train.end, type=type,
 resampling=c("boot","cv","repeatedcv")[1],preProcess = "center")

testData1 <- window(output$dataused,start="2019-01-01",end=end(output$dataused))
P1 <- iForecast(Model=output,Type="static",newdata=testData1)
P2 <- iForecast(Model=output,Type="dynamic",n.ahead=nrow(testData1))

tail(cbind(testData1[,1],P1,P2))

#Case 2. High frequency
#head(ES_15m)
#head(ES_Daily)
#dep <- ES_15m #SP500 15-minute realized absolute variance
#ind <- NULL
#train.end <- as.character(rownames(dep))[as.integer(nrow(dep)*0.9)]

#models<-c("svm","rf","rpart","gamboost","BstLm","bstSm","blackboost")[1]
#type<-c("none","trend","season","both")[1]
# output <- tts.caret(y=dep, x=ind, arOrder=c(3,5), xregOrder=c(0,2,4),
# method=models, tuneLength =10, train.end, type=type,
# resampling=c("boot","cv","repeatedcv")[2],preProcess = "center")
#testData1<-window(output$dataused,start="2009-01-01",end=end(output$dataused))
#P1<-iForecast(Model=output,Type="static",newdata=testData1)
#P2<-iForecast(Model=output,Type="dynamic",n.ahead=nrow(testData1))

Estimate Vector AutoregRessive model by tts.caret

Description

It estimate VAR model by tts.caret, and generates an object list for multistep forecasts.

Usage


tts.var(
  data,
  p,
  method,
  train.end,
  type,
  trace=TRUE)

Arguments

data

The time series object of the VAR dataset, for example, timeSeries,xts, or zoo. y can be either binary or continuous. Date format must be "

p

The lag order as in VAR(p).

method

The train_model_list of caret. While using this, make sure that the method allows regression. Methods in c("svm","rf","rpart","gamboost","BstLm","bstSm","blackboost") are feasible.

train.end

The end date of training data, must be specificed.The default dates of train.start and test.end are the start and the end of input data; and the test.start is the 1-period next of train.end.

type

The time dummies variables. We have four selection:
"none"=no other variables,
"trend"=inclusion of time dummy,
"season"=inclusion of seasonal dummies,
"both"=inclusion of both trend and season. No default.

trace

Whether to print the looping information. The defaut is TRUE.

Details

This function calls tts.caret of package to execute VAR estimation.

Value

output

Output list object generated.

method

The method used.

type

Type of time dummies used, inherited from type of tts.var

data

The complete data structure

Author(s)

Ho Tsung-wu <tsungwu@ntnu.edu.tw>, College of Management, National Taiwan Normal University.

Examples

data(macrodata)
y=timeSeries::as.timeSeries(macrodata[,-1])
VLD=window(y,start="2019-01-01",end=end(y))
#OUT1=tts.var(data=y,
#             p=3,
#             method="enet",
#             train.end="2018-12-01",
#             type=c("none","trend","season","both")[1])

#fcst_ml=iForecast.var(OUT1, n.ahead=nrow(VLD))