Load data from Layman Brothers and define variables to train

src-load-split-data.r

log_debug('Load data')
layman_brothers_url = 
  "https://raw.githubusercontent.com/fclesio/learning-space/master/Datasets/02%20-%20Classification/default_credit_card.csv"

layman_brothers.hex = h2o.importFile(path = layman_brothers_url,
                                     destination_frame = "layman_brothers.hex")
log_debug('Data loaded')

log_debug('Transform default variable to factor')
layman_brothers.hex$DEFAULT = as.factor(layman_brothers.hex$DEFAULT)


log_debug('Construct test and train sets using sampling')
layman_brothers.split <- h2o.splitFrame(data = layman_brothers.hex,
                                        ratios = 0.90, seed =42)

layman_brothers.train <- layman_brothers.split[[1]]
layman_brothers.test <- layman_brothers.split[[2]]

qty_samples_train <- nrow(layman_brothers.train)
qty_samples_test <- nrow(layman_brothers.test)

log_debug('Training set with {qty_samples_train} records')
log_debug('Test set with {qty_samples_test} records')

log_debug('Set predictor and response variables')
y = "DEFAULT"

x = c(
  "LIMIT_BAL"
  ,"SEX"
  ,"EDUCATION"
  ,"MARRIAGE"
  ,"AGE"
  ,"PAY_0"
  ,"PAY_2"
  ,"PAY_3"
  ,"PAY_4"
  ,"PAY_5"
  ,"PAY_6"
  ,"BILL_AMT1"
  ,"BILL_AMT2"
  ,"BILL_AMT3"
  ,"BILL_AMT4"
  ,"BILL_AMT5"
  ,"BILL_AMT6"
  ,"PAY_AMT1"
  ,"PAY_AMT2"
  ,"PAY_AMT3"
  ,"PAY_AMT4"
  ,"PAY_AMT5"
  ,"PAY_AMT6")