Description Usage Arguments Value Author(s) Examples
To get (back) to the overview of all steps and functions use this link:
a.a.main
This is a wrapper function for a random forest classifier.
It executes these functions:
d.b.step1
d.c.step2
d.d.evaluation
d.e.plotImportanceColorHist
1 2 | d.a.randomForest.start(block, feature1, saveFile, classLabel, numTrees,
feature2 = NULL, mtry = NULL, nodesize = NULL, cutoff = NULL)
|
block |
A string that holds a path to a cross-validation block file |
feature1 |
A string that holds a path to a feature data file (hog or colorHist) |
saveFile |
A string that holds a path to which the result should be stored as .rda file. |
classLabel |
A string that holds a path to a file with classification labels for the feature data |
numTrees |
A number that defines the number of trees randomForest should use (see package randomForest) |
feature2 |
A string that holds a path to a second feature data file (the type that was not used in |
mtry |
A number that defines the number of variables randomly sampled as candidates at each split of randomForest (see package randomForest). This parameter is optional |
nodesize |
A number that defines the minimum size of terminal nodes for randomForest (see package randomForest). This parameter is optional |
cutoff |
A vector of length equal to number of classes that defines the ratio of proportion of votes to cutoff (see package randomForest). This parameter is optional |
A table that holds the pairs of correct and not correct predicted images, the number of correct predicted images and the accuracy percentage
Vitali Friesen, Colin Juers, Tassilo Tobollik
1 2 3 | \code{result <- d.a.randomForest.start("data/blocks2677IMG.rda", "data/hog_original_8_9_complete.rda",
"data/colorHistRGBNorm_original_255buckets_hog_original_8_9_nodes7_rf100_result.rda", "data/classesOrig.rda",
100, feature2 = "data/colorHistOriginalRGBNorm255Buckets.rda",nodesize = 7)}
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.