ordDataGen function

Artificial data for testing ordEval algorithms

The generator produces ordinal data simulating different profiles of attributes: basic, performance, excitement and irrelevant.

ordDataGen(noInst, classNoise=0)

Arguments

• noInst: Number of instances to generate.
• classNoise: Proportion of randomly determined values in the class variable.

Details

Problem is described by six important and two irrelevant features. The important features correspond to different feature types from the marketing theory: two basic features ($B_{weak}$ and $B_{strong}$), two performance features ($P_{weak}$

and $P_{strong}$), two excitement features ($E_{weak}$ and $E_{strong}$), and two irrelevant features ($I_{uniform}$ and $I_{normal}$). The values of all features are randomly generated integer values from 1 to 5, indicating for example score assigned to each of the features by the survey's respondent. The dependent variable for each instance (class) is the sum of its features' effects, which we scale to the uniform distribution of integers 1-5, indicating, for example, an overall score assigned by the respondent.

Returns

The method returns a data.frame with noInst rows and 9 columns. Range of values of the attributes and class are integers in [1,5]

Author(s)

Marko Robnik-Sikonja

See Also

classDataGen, regDataGen, ordEval,

Examples

#prepare a data set
dat <- ordDataGen(200)

# evaluate ordered features with ordEval
est <- ordEval(class ~ ., dat, ordEvalNoRandomNormalizers=100)
# print(est)  
plot(est)