`trainExplore()` finds the best decision rule for the given data set based using EXPLORE.
Usage
trainExplore(
train_data = NULL,
settings_path = NULL,
output_path = tempdir(),
file_name = "train_data",
OutputFile = NULL,
StartRulelength = 1,
EndRulelength = 3,
OperatorMethod = "EXHAUSTIVE",
CutoffMethod = "RVAC",
ClassFeature = "'class'",
PositiveClass = "'Iris-versicolor'",
FeatureInclude = "",
Maximize = "ACCURACY",
Accuracy = 0,
BalancedAccuracy = 0,
Specificity = 0,
PrintSettings = TRUE,
PrintPerformance = TRUE,
Subsumption = TRUE,
BranchBound = TRUE,
Parallel = FALSE
)Arguments
- train_data
Train data
- settings_path
A string declaring the path to the settings
- output_path
A string declaring the path to the settings
- file_name
A string declaring the the path to the file name
- OutputFile
A string declaring the path to the output file
- StartRulelength
Positive integer
- EndRulelength
Positive integer
- OperatorMethod
One of list with strings, e.g. list = "EXHAUSTIVE", ...
- CutoffMethod
One of list with strings, list = "RVAC", ...
- ClassFeature
String, should be name of one of columns in data train. Always provided by the user. The string should be enclused in single quotation marks, e.g. 'class'
- PositiveClass
1 or string (?) (should be one of elements of column 'ClassFeature' in data train). Always provided by the user. The string should be enclused in single quotation marks, e.g. 'class'
- FeatureInclude
Empty or string (should be name of one of columns in data train)
- Maximize
One of list with strings, list = "ACCURACY", "SENSITIVITY", "SPECIFICITY", ...
- Accuracy
Float 0-0.999 -> default = 0 (if 0, make empty = computationally more beneficial)
- BalancedAccuracy
Float 0-0.999 -> default = 0 (if 0, make empty = computationally more beneficial)
- Specificity
float 0-0.999, default = 0
- PrintSettings
True or False
- PrintPerformance
True or False
- Subsumption
True or False
- BranchBound
True or False
- Parallel
True or False