Description Usage Arguments Value Author(s) References Examples
An implementation of the mRMR feature selection freamework for the purpose of DISESOR project. In this version the random probe test is used as the main stopping criteria.
1 2 3 | mRMRfs(dataT, target, dependencyF = corrDependency,
randomnessTest = permutationTest, Nprobes = 1000,
allowedRandomness = 0.01, nMax = 20, nCores = 1, ...)
|
dataT |
a data table in |
target |
a vector of target values for data in |
dependencyF |
a function for computing dependencies between attributes and between
attributes and the decisions. The default is an absolute value of
Pearson's correlation (function |
randomnessTest |
a function implementing a randomness test used as a stopping criteria.
The default ( |
Nprobes |
an integer specifying the number of probes to use in estimation of
attribute irrelevance (for stopping criteria). The default is |
allowedRandomness |
a numeric value specifying allowed attribute irrelevance
probability. The default is |
nMax |
an integer specifying maximal number of features that can be returned by
the function. The default is |
nCores |
an integer specifying the number of available processor cores for parallel
computations using forking. The default is |
... |
optional arguments (currently omitted). |
an integer vector representing indexes of attributes from the selected subset.
Andrzej Janusz
Hanchuan Peng, Fuhui Long, and Chris Ding. Feature selection based on mutual information: Criteria of max-dependency, max-relevance, and min-redundancy. IEEE Trans. Pattern Anal. Mach. Intell., 27(8):1226–1238
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 | #############################################
data(methaneSampleData)
## an experiment on a sample from the data used in a data mining competition -
## IJCRS'15 Data Challenge: Mining Data from Coal Mines
## (https://knowledgepit.fedcsis.org/contest/view.php?id=109).
## The whole data set can be downloaded from the competition web page.
mrmrAttrs = mRMRfs(dataT = methaneData$methaneTraining,
target = methaneData$methaneTrainingLabels[, as.integer(V2 == 'warning')],
dependencyF = corrDependency)
mrmrAttrs
regModel = glm(targets ~.,
cbind(methaneData$methaneTraining[, mrmrAttrs, with = FALSE],
targets = methaneData$methaneTrainingLabels[, as.integer(V2 == 'warning')]),
family = gaussian(link = "identity"))
preds = predict(regModel, methaneData$methaneTest, type = "response")
caTools::colAUC(preds, methaneData$methaneTestLabels[, V2])
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.