Problems with Date

[alexeks]

Modeltimeh2o error is

Error in .h2o.doSafeREST(h2oRestApiVersion = h2oRestApiVersion, urlSuffix = page, :
ERROR MESSAGE:
Provided column type POSIXct is unknown. Cannot proceed with parse due to invalid argument.
If the Date is passed as a date (without minutes ann seconds) there is no error message

h2o code itself is running with minutes and seconds with no problems.
Both codes are below. The data file is attached.

library(modeltime)
library(modeltime.h2o)
library(modeltime.resample)
library(modeltime.ensemble)
library(timetk)
library(tidymodels)
library(tidyverse)
library(dplyr)
library(tidyr)
library(lubridate)
h20.csv

#h2o code without modeltime h2o
h2o.init(
nthreads = -1,
ip = 'localhost',
port = 54321
)

h2odf1<-h2o.importFile("C:/City of Chico/h20.csv")

aml <- h2o.automl(x = "Date", y = "total",
training_frame = h2odf1,
max_models = 20,
seed = 1)

lb <- aml@leaderboard
print(lb, n = nrow(lb))
#modeltime h2o code#################
h2o.init(
nthreads = -1,
ip = 'localhost',
port = 54321
)

model_spec <- automl_reg(mode = 'regression') %>%
set_engine(
engine = 'h2o',
max_runtime_secs = 5,
max_runtime_secs_per_model = 3,
max_models = 3,
nfolds = 5,
exclude_algos = c("DeepLearning"),
verbosity = NULL,
seed = 786
)

model_fitted1 <- model_spec %>%
fit(total ~ Date,data=as.data.frame(h2odf1))

[Matt Dancho]

Will take a look and let you know.

[Muhammad Ichsan]

Modeltimeh2o error is

Error in .h2o.doSafeREST(h2oRestApiVersion = h2oRestApiVersion, urlSuffix = page, : ERROR MESSAGE: Provided column type POSIXct is unknown. Cannot proceed with parse due to invalid argument.
If the Date is passed as a date (without minutes ann seconds) there is no error message

h2o code itself is running with minutes and seconds with no problems.

Both codes are below. The data file is attached.
library(modeltime)
library(modeltime.h2o)
library(modeltime.resample)
library(modeltime.ensemble)
library(timetk)
library(tidymodels)
library(tidyverse)
library(dplyr)
library(tidyr)
library(lubridate)
h20.csv

#h2o code without modeltime h2o
h2o.init(
nthreads = -1,
ip = 'localhost',
port = 54321
)

h2odf1<-h2o.importFile("C:/City of Chico/h20.csv")

aml <- h2o.automl(x = "Date", y = "total",
training_frame = h2odf1,
max_models = 20,
seed = 1)

lb <- aml@leaderboard
print(lb, n = nrow(lb))
#modeltime h2o code#################
h2o.init(
nthreads = -1,
ip = 'localhost',
port = 54321
)

model_spec <- automl_reg(mode = 'regression') %>%
set_engine(
engine = 'h2o',
max_runtime_secs = 5,
max_runtime_secs_per_model = 3,
max_models = 3,
nfolds = 5,
exclude_algos = c("DeepLearning"),
verbosity = NULL,
seed = 786
)

model_fitted1 <- model_spec %>%
fit(total ~ Date,data=as.data.frame(h2odf1))

you can try
h2odf1_rev <- h2odf1 %>% mutate(Date = as.Date(Date))

then, you use that dataframe in this syntax
model_fitted1 <- model_spec %>% fit(total ~ Date, data=h2odf1_rev)