inst/tests/test-pred-newdata-checks.R
In appling/rloadest_tests: Test patches to rloadest
test.inoutputs.path="../loadest_examples/app" # for calls from test()
#test.inoutputs.path="inst/loadest_examples/app" # for running these lines manually
for(withfix in c(FALSE, TRUE)) {
if(withfix) {
# To see if the patch worked, go into dev_mode and reload rloadest as the version in pangia/rloadest/pred-newdata-checks
dev_mode(on=TRUE)
unload(inst("rloadest"))
install_github(repo="pangia/rloadest", ref="pred-newdata-checks")
}
test_that(paste(if(withfix) "with" else "without","fix, rloadest returns informative errors when newdata is missing columns in predLoad"), {
### Setup ###
library("rloadest")
LOADEST.inputs <- read_LOADEST_inputs(inputs.path=paste0(test.inoutputs.path,"1"))
# Get & munge the calibration data
calib.inp <- LOADEST.inputs$calib.inp
names(calib.inp)[which(names(calib.inp)=="CCONC")] <- "Phosphorus"
calib.inp$DATES <- as.Date(strptime(as.character(calib.inp$CDATE), format="%Y%m%d"))
calib.inp$TIMES <- as.character(calib.inp$CTIME)
calib.inp$FLOW <- calib.inp$CFLOW
# We've now duplicated the rloadest data from the LOADEST data:
expect_that(calib.inp[c("DATES","TIMES","FLOW","Phosphorus")], equals(app1.calib))
# Fit a loadReg model
app1_lr <- loadReg(
formula=Phosphorus ~ model(1),
data=calib.inp[c("DATES","FLOW","Phosphorus")],
flow="FLOW",
dates="DATES",
conc.units="mg/L",
station="Illinois River at Marseilles, Ill.")
expect_that(class(app1_lr) == "loadReg", is_true())
# Get & munge the estimation data
est.inp <- LOADEST.inputs$est.inp
est.inp$DATES <- as.Date(strptime(as.character(est.inp$EDATE), format="%Y%m%d"))
est.inp$TIMES <- as.character(est.inp$ETIME)
est.inp$FLOW <- est.inp$EFLOW
est.inp$GROUPING <- rep(1:4, each=24)
# Again, we've duplicated the rloadest data from the LOADEST data:
expect_that(est.inp[c("DATES","TIMES","FLOW")], equals(app1.calib[-which(names(app1.calib)=="Phosphorus")]))
### call predLoad - ERRORS ###
# This call to predLoad is successful
expect_that(predLoad(newdata=est.inp[c("DATES","FLOW")], fit=app1_lr, load.units="tons", by="unit"), is_a("data.frame"))
expect_that(nrow(predLoad(newdata=est.inp[c("DATES","FLOW")], fit=app1_lr, load.units="tons", by="unit")), equals(nrow(est.inp)))
# This call to predLoad fails because app1_lr implies that est.inp has a FLOW column, but est.inp actually has EFLOW
# Failure is fine. However, the error is unhelpful and differs depending on the value of the 'by' argument.
# What we get:
if(!withfix) {
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="unit"), throws_error("arguments imply differing number of rows"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="day"), throws_error("arguments imply differing number of rows"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="month"), throws_error("arguments must have same length"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="water year"), throws_error("arguments must have same length"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="calendar year"), throws_error("arguments must have same length"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], load.units="tons", by="GROUPING"), throws_error("arguments must have same length"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="total"), throws_error("subscript out of bounds"))
}
### More elegant failures ###
# What would be more helpful would be an error that tells you how to fix the problem before it creates others
new_error <- "newdata has missing, misnamed, or malformed predictor columns"
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="unit"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="day"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="month"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="water year"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="calendar year"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], load.units="tons", by="GROUPING"), throws_error(new_error))
expect_that(predLoad(newdata=est.inp[c("DATES","EFLOW")], fit=app1_lr, load.units="tons", by="total"), throws_error(new_error))
# One solution to generate errors like this would be to add error handling at
# line 104, just after model.inp has been set by either model.matrix or
# setXLDat. For example, I've tried adding the following at predLoad:line104:
# if (nrow(model.inp) == 0) stop("one or more expected columns in newdata were not found")
# This is effective at stopping the above calls with that error message for
# any value of 'by'. A few more lines of code at that location could generate
# an even more explicit error message that explains which column (or other
# mismatch) created the problem.
### Proximate causes of error ###
# I've traced these problems to the following errors.
# by="unit" or by="day" throw errors at the point after estlday where they call
# retval <- data.frame(Date = newdata[[dates]], Flow = Flow,
# Flux = Flux, Std.Err = Std.Err, SEP = SEP, L95 = L95,
# U95 = U95)
# Because of the vectors passed to data.frame, all but newdata[[dates]] have length 0.
# "estlday" has returned values of length 0 because model.inp has 0 rows.
# model.inp really should have nrow(est.inp) rows, but it doesn't because
# newdata$FLOW was not available, and this problem hasn't been caught by this
# point in the code.
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="unit"), throws_error("arguments imply differing number of rows"))
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="day"), throws_error("arguments imply differing number of rows"))
# by="month","water year","calendar year", or an arbitrary column names gives
# problems at the call to tapply(seq(nrow(model.inp)), newdata[[by]], ...) on
# line 360. seq(nrow(model.inp)) has length 2, while newdata has length 96.
# They need to have the same length. This could be caught by comparing the
# lengths of these two objects, but it seems that a better fix would be to
# make sure that model.inp and newdata have the same nrow, and to run that
# check much earlier in the code (where I suggested putting the stop() call)
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="month"), throws_error("arguments must have same length"))
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="water year"), throws_error("arguments must have same length"))
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="calendar year"), throws_error("arguments must have same length"))
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], load.units="tons", by="GROUPING"), throws_error("arguments must have same length"))
# by='total' fails at line 334, xlpred=model.inp[KDays,], because model.inp
# has 0 rows and cannot be indexed by KDays=c(1,0). The main problem is that
# model.inp has 0 rows when it should have nrow(est.inp) rows. A secondary,
# proximate problem is that KDays is c(1,0) because of the call KDays <-
# seq(nrow(model.inp)) in line 305, because seq(0) gives c(1,0).
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="total"), throws_error("subscript out of bounds"))
# Because all three of these errors stem from the same fundamental problem, I
# think it makes the most sense to prevent these errors through the above
# suggestion to add a check for nrow(model.inp)==0 right after building
# model.inp.
})
test_that(paste(if(withfix) "with" else "without","fix, rloadest returns informative errors when newdata is missing columns in predConc"), {
### Setup ###
library("rloadest")
LOADEST.inputs <- read_LOADEST_inputs(inputs.path=paste0(test.inoutputs.path,"1"))
# Get & munge the calibration data
calib.inp <- LOADEST.inputs$calib.inp
names(calib.inp)[which(names(calib.inp)=="CCONC")] <- "Phosphorus"
calib.inp$DATES <- as.Date(strptime(as.character(calib.inp$CDATE), format="%Y%m%d"))
calib.inp$TIMES <- as.character(calib.inp$CTIME)
calib.inp$FLOW <- calib.inp$CFLOW
# We've now duplicated the rloadest data from the LOADEST data:
expect_that(calib.inp[c("DATES","TIMES","FLOW","Phosphorus")], equals(app1.calib))
# Fit a loadReg model
app1_lr <- loadReg(
formula=Phosphorus ~ model(1),
data=calib.inp[c("DATES","FLOW","Phosphorus")],
flow="FLOW",
dates="DATES",
conc.units="mg/L",
station="Illinois River at Marseilles, Ill.")
expect_that(class(app1_lr) == "loadReg", is_true())
# Get & munge the estimation data
est.inp <- LOADEST.inputs$est.inp
est.inp$DATES <- as.Date(strptime(as.character(est.inp$EDATE), format="%Y%m%d"))
est.inp$TIMES <- as.character(est.inp$ETIME)
est.inp$FLOW <- est.inp$EFLOW
est.inp$GROUPING <- rep(1:4, each=24)
# Again, we've duplicated the rloadest data from the LOADEST data:
expect_that(est.inp[c("DATES","TIMES","FLOW")], equals(app1.calib[-which(names(app1.calib)=="Phosphorus")]))
### call predConc - ERRORS ###
# This call to predConc is successful
expect_that(predConc(newdata=est.inp[c("DATES","FLOW")], fit=app1_lr, by="unit"), is_a("data.frame"))
expect_that(nrow(predConc(newdata=est.inp[c("DATES","FLOW")], fit=app1_lr, by="unit")), equals(nrow(est.inp)))
# This call to predConc fails because app1_lr implies that est.inp has a FLOW column, but est.inp actually has EFLOW
# Failure is fine. However, the error is unhelpful and differs depending on the value of the 'by' argument.
# What we get:
if(!withfix) {
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="unit"), throws_error("arguments imply differing number of rows"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="day"), throws_error("arguments imply differing number of rows"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="month"), throws_error("arguments must have same length"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="water year"), throws_error("arguments must have same length"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="calendar year"), throws_error("arguments must have same length"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], by="GROUPING"), throws_error("arguments must have same length"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="total"), throws_error("subscript out of bounds"))
}
### More elegant failures ###
# What would be more helpful would be an error that tells you how to fix the problem before it creates others
new_error <- "newdata has missing, misnamed, or malformed predictor columns"
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="unit"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="day"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="month"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="water year"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="calendar year"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], by="GROUPING"), throws_error(new_error))
expect_that(predConc(newdata=est.inp[c("DATES","EFLOW")], fit=app1_lr, by="total"), throws_error(new_error))
})
if(withfix) {
# Leave dev_mode and reload rloadest from the usual directory
unload(inst("rloadest"))
dev_mode(on=FALSE)
library("rloadest")
}
}
appling/rloadest_tests documentation built on May 10, 2019, 12:44 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
test.inoutputs.path="../loadest_examples/app" # for calls from test()
#test.inoutputs.path="inst/loadest_examples/app" # for running these lines manually
for(withfix in c(FALSE, TRUE)) {
if(withfix) {
# To see if the patch worked, go into dev_mode and reload rloadest as the version in pangia/rloadest/pred-newdata-checks
dev_mode(on=TRUE)
unload(inst("rloadest"))
install_github(repo="pangia/rloadest", ref="pred-newdata-checks")
}
test_that(paste(if(withfix) "with" else "without","fix, rloadest returns informative errors when newdata is missing columns in predLoad"), {
### Setup ###
library("rloadest")
LOADEST.inputs <- read_LOADEST_inputs(inputs.path=paste0(test.inoutputs.path,"1"))
# Get & munge the calibration data
calib.inp <- LOADEST.inputs$calib.inp
names(calib.inp)[which(names(calib.inp)=="CCONC")] <- "Phosphorus"
calib.inp$DATES <- as.Date(strptime(as.character(calib.inp$CDATE), format="%Y%m%d"))
calib.inp$TIMES <- as.character(calib.inp$CTIME)
calib.inp$FLOW <- calib.inp$CFLOW
# We've now duplicated the rloadest data from the LOADEST data:
expect_that(calib.inp[c("DATES","TIMES","FLOW","Phosphorus")], equals(app1.calib))
# Fit a loadReg model
app1_lr <- loadReg(
formula=Phosphorus ~ model(1),
data=calib.inp[c("DATES","FLOW","Phosphorus")],
flow="FLOW",
dates="DATES",
conc.units="mg/L",
station="Illinois River at Marseilles, Ill.")
expect_that(class(app1_lr) == "loadReg", is_true())
# Get & munge the estimation data
est.inp <- LOADEST.inputs$est.inp
est.inp$DATES <- as.Date(strptime(as.character(est.inp$EDATE), format="%Y%m%d"))
est.inp$TIMES <- as.character(est.inp$ETIME)
est.inp$FLOW <- est.inp$EFLOW
est.inp$GROUPING <- rep(1:4, each=24)
# Again, we've duplicated the rloadest data from the LOADEST data:
expect_that(est.inp[c("DATES","TIMES","FLOW")], equals(app1.calib[-which(names(app1.calib)=="Phosphorus")]))
### call predLoad - ERRORS ###
# This call to predLoad is successful
expect_that(predLoad(newdata=est.inp[c("DATES","FLOW")], fit=app1_lr, load.units="tons", by="unit"), is_a("data.frame"))
expect_that(nrow(predLoad(newdata=est.inp[c("DATES","FLOW")], fit=app1_lr, load.units="tons", by="unit")), equals(nrow(est.inp)))
# This call to predLoad fails because app1_lr implies that est.inp has a FLOW column, but est.inp actually has EFLOW
# Failure is fine. However, the error is unhelpful and differs depending on the value of the 'by' argument.
# What we get:
if(!withfix) {
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="unit"), throws_error("arguments imply differing number of rows"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="day"), throws_error("arguments imply differing number of rows"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="month"), throws_error("arguments must have same length"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="water year"), throws_error("arguments must have same length"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="calendar year"), throws_error("arguments must have same length"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], load.units="tons", by="GROUPING"), throws_error("arguments must have same length"))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="total"), throws_error("subscript out of bounds"))
}
### More elegant failures ###
# What would be more helpful would be an error that tells you how to fix the problem before it creates others
new_error <- "newdata has missing, misnamed, or malformed predictor columns"
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="unit"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="day"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="month"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="water year"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="calendar year"), throws_error(new_error))
expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], load.units="tons", by="GROUPING"), throws_error(new_error))
expect_that(predLoad(newdata=est.inp[c("DATES","EFLOW")], fit=app1_lr, load.units="tons", by="total"), throws_error(new_error))
# One solution to generate errors like this would be to add error handling at
# line 104, just after model.inp has been set by either model.matrix or
# setXLDat. For example, I've tried adding the following at predLoad:line104:
# if (nrow(model.inp) == 0) stop("one or more expected columns in newdata were not found")
# This is effective at stopping the above calls with that error message for
# any value of 'by'. A few more lines of code at that location could generate
# an even more explicit error message that explains which column (or other
# mismatch) created the problem.
### Proximate causes of error ###
# I've traced these problems to the following errors.
# by="unit" or by="day" throw errors at the point after estlday where they call
# retval <- data.frame(Date = newdata[[dates]], Flow = Flow,
# Flux = Flux, Std.Err = Std.Err, SEP = SEP, L95 = L95,
# U95 = U95)
# Because of the vectors passed to data.frame, all but newdata[[dates]] have length 0.
# "estlday" has returned values of length 0 because model.inp has 0 rows.
# model.inp really should have nrow(est.inp) rows, but it doesn't because
# newdata$FLOW was not available, and this problem hasn't been caught by this
# point in the code.
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="unit"), throws_error("arguments imply differing number of rows"))
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="day"), throws_error("arguments imply differing number of rows"))
# by="month","water year","calendar year", or an arbitrary column names gives
# problems at the call to tapply(seq(nrow(model.inp)), newdata[[by]], ...) on
# line 360. seq(nrow(model.inp)) has length 2, while newdata has length 96.
# They need to have the same length. This could be caught by comparing the
# lengths of these two objects, but it seems that a better fix would be to
# make sure that model.inp and newdata have the same nrow, and to run that
# check much earlier in the code (where I suggested putting the stop() call)
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="month"), throws_error("arguments must have same length"))
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="water year"), throws_error("arguments must have same length"))
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="calendar year"), throws_error("arguments must have same length"))
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], load.units="tons", by="GROUPING"), throws_error("arguments must have same length"))
# by='total' fails at line 334, xlpred=model.inp[KDays,], because model.inp
# has 0 rows and cannot be indexed by KDays=c(1,0). The main problem is that
# model.inp has 0 rows when it should have nrow(est.inp) rows. A secondary,
# proximate problem is that KDays is c(1,0) because of the call KDays <-
# seq(nrow(model.inp)) in line 305, because seq(0) gives c(1,0).
#expect_that(predLoad(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], load.units="tons", by="total"), throws_error("subscript out of bounds"))
# Because all three of these errors stem from the same fundamental problem, I
# think it makes the most sense to prevent these errors through the above
# suggestion to add a check for nrow(model.inp)==0 right after building
# model.inp.
})
test_that(paste(if(withfix) "with" else "without","fix, rloadest returns informative errors when newdata is missing columns in predConc"), {
### Setup ###
library("rloadest")
LOADEST.inputs <- read_LOADEST_inputs(inputs.path=paste0(test.inoutputs.path,"1"))
# Get & munge the calibration data
calib.inp <- LOADEST.inputs$calib.inp
names(calib.inp)[which(names(calib.inp)=="CCONC")] <- "Phosphorus"
calib.inp$DATES <- as.Date(strptime(as.character(calib.inp$CDATE), format="%Y%m%d"))
calib.inp$TIMES <- as.character(calib.inp$CTIME)
calib.inp$FLOW <- calib.inp$CFLOW
# We've now duplicated the rloadest data from the LOADEST data:
expect_that(calib.inp[c("DATES","TIMES","FLOW","Phosphorus")], equals(app1.calib))
# Fit a loadReg model
app1_lr <- loadReg(
formula=Phosphorus ~ model(1),
data=calib.inp[c("DATES","FLOW","Phosphorus")],
flow="FLOW",
dates="DATES",
conc.units="mg/L",
station="Illinois River at Marseilles, Ill.")
expect_that(class(app1_lr) == "loadReg", is_true())
# Get & munge the estimation data
est.inp <- LOADEST.inputs$est.inp
est.inp$DATES <- as.Date(strptime(as.character(est.inp$EDATE), format="%Y%m%d"))
est.inp$TIMES <- as.character(est.inp$ETIME)
est.inp$FLOW <- est.inp$EFLOW
est.inp$GROUPING <- rep(1:4, each=24)
# Again, we've duplicated the rloadest data from the LOADEST data:
expect_that(est.inp[c("DATES","TIMES","FLOW")], equals(app1.calib[-which(names(app1.calib)=="Phosphorus")]))
### call predConc - ERRORS ###
# This call to predConc is successful
expect_that(predConc(newdata=est.inp[c("DATES","FLOW")], fit=app1_lr, by="unit"), is_a("data.frame"))
expect_that(nrow(predConc(newdata=est.inp[c("DATES","FLOW")], fit=app1_lr, by="unit")), equals(nrow(est.inp)))
# This call to predConc fails because app1_lr implies that est.inp has a FLOW column, but est.inp actually has EFLOW
# Failure is fine. However, the error is unhelpful and differs depending on the value of the 'by' argument.
# What we get:
if(!withfix) {
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="unit"), throws_error("arguments imply differing number of rows"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="day"), throws_error("arguments imply differing number of rows"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="month"), throws_error("arguments must have same length"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="water year"), throws_error("arguments must have same length"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="calendar year"), throws_error("arguments must have same length"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], by="GROUPING"), throws_error("arguments must have same length"))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="total"), throws_error("subscript out of bounds"))
}
### More elegant failures ###
# What would be more helpful would be an error that tells you how to fix the problem before it creates others
new_error <- "newdata has missing, misnamed, or malformed predictor columns"
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="unit"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="day"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="month"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="water year"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW")], by="calendar year"), throws_error(new_error))
expect_that(predConc(fit=app1_lr, newdata=est.inp[c("DATES","EFLOW","GROUPING")], by="GROUPING"), throws_error(new_error))
expect_that(predConc(newdata=est.inp[c("DATES","EFLOW")], fit=app1_lr, by="total"), throws_error(new_error))
})
if(withfix) {
# Leave dev_mode and reload rloadest from the usual directory
unload(inst("rloadest"))
dev_mode(on=FALSE)
library("rloadest")
}
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.