predict: Prediction of a 'PrecipitationOccurrenceModel' model object

In ecor/RGENERATEPREC: Tools to Generate Daily-Precipitation Time Series

Prediction of a PrecipitationOccurrenceModel model object

Description

It is a wrapper of predict.glm method for the a PrecipitationOccurrenceModel model object S3 class.

Usage

## S3 method for class 'PrecipitationOccurrenceModel'
predict(
  object,
  newdata = NULL,
  type = "response",
  previous = NULL,
  endogenous = NULL,
  ...
)

## S3 method for class 'PrecipitationOccurrenceMultiSiteModel'
predict(object, ...)

## S3 method for class 'PrecipitationAmountModel'
predict(
  object,
  newdata = NULL,
  origin_newdata = NA,
  precipitation.value.random.generation = FALSE,
  ...
)

Arguments

object	model returned by PrecipitationOccurrenceModel
newdata	predictor or exogenous variables
type	see predict.glm. Default is "response". See predict.glm.
previous	logical vector containing previously occurred states.
endogenous	String vector containing the name of the endogenous variables. It is used if the endogenous variables are more than one, otherwise is set NULL(Default).
...	further arguments
origin_newdata	character string containing the date corresponding the first row of newdata
precipitation.value.random.generation	logical value. If it is FALSE (Default) the method predict.PrecipitationAmountModel returns conditioned random values, otherwise these values are converted to precipitation values through their observed non-parametric distributions.

Value

A vector or a data frame reporting predicted time series for each station.

See Also

predict.glm,PrecipitationOccurrenceModel

predict.glm,predict.glm,PrecipitationOccurrenceModel,PrecipitationAmountModel

Examples

library(RGENERATEPREC)

data(trentino)

year_min <- 1961
year_max <- 1990

period <- PRECIPITATION$year>=year_min & PRECIPITATION$year<=year_max
period_temp <- TEMPERATURE_MAX$year>=year_min & TEMPERATURE_MAX$year<=year_max

prec_mes <- PRECIPITATION[period,]
Tx_mes <- TEMPERATURE_MAX[period_temp,]
Tn_mes <- TEMPERATURE_MIN[period_temp,]
accepted <- array(TRUE,length(names(prec_mes)))
names(accepted) <- names(prec_mes)
for (it in names(prec_mes)) {
	acc <- TRUE
	acc <- (length(which(!is.na(Tx_mes[,it])))==length(Tx_mes[,it]))
	acc <- (length(which(!is.na(Tn_mes[,it])))==length(Tn_mes[,it])) & acc
	accepted[it]  <- (length(which(!is.na(prec_mes[,it])))==length(prec_mes[,it])) & acc
	
}

valmin <- 1.0
prec_mes <- prec_mes[,accepted]



Tx_mes <- Tx_mes[,accepted]
Tn_mes <- Tn_mes[,accepted]
origin <- paste(year_min,1,1,sep="-")


prec_occurrence_mes <- prec_mes>=valmin

station <- names(prec_mes)[!(names(prec_mes) %in% c("day","month","year"))]
it <- station[2]
vect <- Tx_mes[,it]-Tn_mes[,it]
months <- factor(prec_mes$month)
model <- PrecipitationOccurrenceModel(x=prec_mes[,it],exogen=vect,monthly.factor=months)

probs <- predict(model)

nday <- 3.0
vect_new <- array(1.0,nday)
months_new <- array(1,nday)
row_test <- 2000:2007
newdata <- model$predictor[row_test,]
probs2 <- predict(model,newdata=newdata)

probs[row_test]==probs2
###





prec_occurrence_mes <- prec_mes>=valmin

station <- names(prec_mes)[!(names(prec_mes) %in% c("day","month","year"))]

station <- station[1:4] ## reduced the dataset!!! 
Tx_mes <- Tx_mes[,station]
Tn_mes <- Tn_mes[,station]

prec_mes <- prec_mes[,station]
exogen <- Tx_mes-Tn_mes
months <- factor(prec_mes$month)


### Not Run 
### Please uncomment the following lines to run them 


model_multisite <- PrecipitationOccurrenceMultiSiteModel(x=prec_mes,
exogen=exogen,origin=origin,multisite_type="wilks")


model_multisite_logit <- PrecipitationOccurrenceMultiSiteModel(x=prec_mes,
exogen=exogen,origin=origin,multisite_type="logit")


probs_multimodel  <- predict(model_multisite_logit)

ecor/RGENERATEPREC documentation built on Feb. 3, 2023, 10:38 a.m.