ecm (version 7.2.0)

ecmpredict: Predict using an ecm object

Description

Takes an ecm object and uses it to predict based on new data. This prediction does the undifferencing required to transform the change in y back to y itself.

Usage

ecmpredict(model, newdata, init)

Value

Numeric predictions on new data based ecm object

Arguments

model

ecm object used to make predictions

newdata

Data frame to on which to predict

init

Initial value(s) for prediction

Details

Since error correction models only model the change in the target variable, an initial value must be specified. Additionally, the 'newdata' parameter should have at least 3 rows of data.

Examples

##Not run

data(Wilshire)

#Rebuilding model1 from ecm example
trn <- Wilshire[Wilshire$date<='2015-12-01',]
xeq <- xtr <- trn[c('CorpProfits', 'FedFundsRate', 'UnempRate')]
model1 <- ecm(trn$Wilshire5000, xeq, xtr)
model2 <- ecm(trn$Wilshire5000, xeq, xtr, linearFitter='earth')

#Use 2015-12-01 and onwards data as test data to predict
tst <- Wilshire[Wilshire$date>='2015-12-01',]

#predict on tst using model1 and initial FedFundsRate
tst$model1Pred <- ecmpredict(model1, tst, tst$Wilshire5000[1])
tst$model2Pred <- ecmpredict(model2, tst, tst$Wilshire5000[1])