RemixAutoML (version 0.4.2)

AutoHurdleScoring: AutoHurdleScoring()

Description

AutoHurdleScoring() can score AutoCatBoostHurdleModel() and AutoXGBoostHurdleModel()

Usage

AutoHurdleScoring(
  TestData = NULL,
  Path = NULL,
  ModelID = NULL,
  ModelClass = "catboost",
  ArgList = NULL,
  ModelList = NULL,
  Threshold = NULL
)

Arguments

TestData

scoring data.table

Path

Supply if ArgList is NULL or ModelList is null.

ModelID

Supply if ArgList is NULL or ModelList is null. Same as used in model training.

ModelClass

Name of model type. "catboost" is currently the only available option

ArgList

Output from the hurdle model

ModelList

Output from the hurdle model

Threshold

NULL to use raw probabilities to predict. Otherwise, supply a threshold

Value

A data.table with the final predicted value, the intermediate model predictions, and your source data

See Also

Other Automated Model Scoring: AutoCatBoostScoring(), AutoH2OMLScoring(), AutoH2OModeler(), AutoXGBoostScoring(), IntermittentDemandScoringDataGenerator()

Examples

# NOT RUN {
# XGBoost----

# Define file path
Path <- "C:/Users/aantico/Documents/Package/GUI_Package"

# Create hurdle data with correlated features
data <- RemixAutoML::FakeDataGenerator(
  Correlation = 0.70,
  N = 25000,
  ID = 3,
  FactorCount = 2L,
  AddDate = TRUE,
  ZIP = 1,
  Classification = FALSE,
  MultiClass = FALSE)

# Define features
Features <- names(data)[!names(data) %chin%
  c("Adrian","IDcol_1","IDcol_2","IDcol_3","DateTime")]

# Build hurdle model
Output <- RemixAutoML::AutoXGBoostHurdleModel(

  # Operationalization args
  TreeMethod = "hist",
  TrainOnFull = FALSE,
  PassInGrid = NULL,

  # Metadata args
  NThreads = max(1L, parallel::detectCores()-2L),
  ModelID = "ModelTest",
  Paths = normalizePath(Path),
  MetaDataPaths = NULL,
  ReturnModelObjects = TRUE,

  # data args
  data,
  ValidationData = NULL,
  TestData = NULL,
  Buckets = c(0),
  TargetColumnName = "Adrian",
  FeatureColNames = Features,
  IDcols = c("IDcol_1","IDcol_2","IDcol_3"),

  # options
  TransformNumericColumns = NULL,
  SplitRatios = c(0.70, 0.20, 0.10),
  SaveModelObjects = TRUE,
  NumOfParDepPlots = 10L,

  # grid tuning args
  GridTune = FALSE,
  grid_eval_metric = "accuracy",
  MaxModelsInGrid = 1L,
  BaselineComparison = "default",
  MaxRunsWithoutNewWinner = 10L,
  MaxRunMinutes = 60L,

  # bandit hyperparameters
  Trees = 100L,
  eta = seq(0.05,0.40,0.05),
  max_depth = seq(4L, 16L, 2L),

  # random hyperparameters
  min_child_weight = seq(1.0, 10.0, 1.0),
  subsample = seq(0.55, 1.0, 0.05),
  colsample_bytree = seq(0.55, 1.0, 0.05))

# Score XGBoost Hurdle Model
HurdleScores <- RemixAutoML::AutoHurdleScoring(
  TestData = data,
  Path = Path,
  ModelID = "ModelTest",
  ModelClass = "xgboost",
  ModelList = NULL,
  ArgList = NULL,
  Threshold = NULL)
# }