## ----setup, include=FALSE-----------------------------------------------------
knitr::opts_chunk$set(echo = TRUE, eval=FALSE)

## -----------------------------------------------------------------------------
# library(cloudml)
# 
# # copy from a local directory to a bucket
# gs_copy("training-data", "gs://quarter-deck-529/training-data")
# 
# # copy from a bucket to a local directory
# gs_copy("gs://quarter-deck-529/training-data", "training-data")

## -----------------------------------------------------------------------------
# # synchronize a bucket and a local directory
# gs_rsync("gs://quarter-deck-529/training-data", "training-data")

## -----------------------------------------------------------------------------
# library(tfdatasets)
# library(cloudml)
# 
# data_dir <- gs_data_dir("gs://mtcars-data")
# mtcars_csv <- file.path(data_dir, "mtcars.csv")
# 
# mtcars_dataset <- csv_dataset(mtcars_csv) %>%
#   dataset_prepare(x = c(mpg, disp), y = cyl)

## -----------------------------------------------------------------------------
# library(cloudml)
# library(readr)
# data_dir <- gs_data_dir_local("gs://quarter-deck-529/training-data")
# train_data <- read_csv(file.path(data_dir, "train.csv"))
# test_data <- read_csv(file.path(data_dir, "test.csv"))

## -----------------------------------------------------------------------------
# train_generator <- flow_images_from_directory(
#   gs_data_dir_local("gs://quarter-deck-529/images/train"),
#   image_data_generator(rescale = 1/255),
#   target_size = c(150, 150),
#   batch_size = 32,
#   class_mode = "binary"
# )

## -----------------------------------------------------------------------------
# library(keras)
# library(cloudml)
# 
# # define a flag for the location of the data directory
# FLAGS <- flags(
#   flag_string("data_dir", "data")
# )
# 
# # determine the location of the directory (during local development this will
# # be the default "data" subdirectory specified in the FLAGS declaration above)
# data_dir <- gs_data_dir_local(FLAGS$data_dir)
# 
# # read the data
# train_data <- read_csv(file.path(FLAGS$data_dir, "train.csv"))
# 

## -----------------------------------------------------------------------------
# gcloud_terminal()