facial-expression-recognition/DecisionTreeClassifier.R at master · Ectelion/facial-expression-recognition · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
# Dependencies
library(rpart)

## Creates Decision Trees based classificator object, containing.
## Includes methods for prediction, cross validation, etc.
## @formula parameter is used to specify labels & features.

initTree <- function(trainingSet, formula = NULL) {
    # Training phase
    tree <- fitDTClassifier(trainingSet, formula)

    tree.predict <- function(data) {
        predict(tree, data, type = "class")
    }

    tree.hitsNum <- function(inputs, trueLabels) {
        hitsNum <- 0
        labels <- c(8)
        falses <- c()
        for (inputId in 1: nrow(inputs)) {
            colsNum <- ncol(inputs)
            probs      <-  predict(tree, inputs[inputId, 1:colsNum])
            max        <- -1
            maxLabelId <-  1
            labelId    <-  1

            for (labelProb in probs) {
                if (labelProb > max) {
                    max <- labelProb
                    maxLabelId <- labelId
                }
                labelId <- labelId + 1
            }
            labels <- c(labels, maxLabelId)
            if (maxLabelId == trueLabels[inputId]) {
                hitsNum <- hitsNum + 1
            } else {
                falses <- c(falses, inputId)
            }
        }
        print(falses)
        hitsNum
    }

    tree.crossValidation <- function(dataSet = trainingSet, K = 10) {
        crossValidation(dataSet, classifierType = "decision_tree", K)
    }

    ## Returns a list representation of the object with methods and properties accessed through indexed keys
    list(classifier = tree, predict = tree.predict, hitsNum = tree.hitsNum, crossValidation = tree.crossValidation)
}

## Fits rpart based Decision Tree classifier

fitDTClassifier <- function(dataSet = loadData(), formula = NULL) {
    if (is.null(formula)) {
        # formulaOptimal <- emotion ~ X52+X87+X89+X92+X101+X102+X117+X118+X120+X123+X125+X1+X23+X40+X24+X99
        # formula <- emotion ~ X52+X87+X89+X92+X101+X102+X117+X118+X120+X123+X125
        # formula <- emotion ~ X52+X87+X89+X92+X101+X102+X117+X118+X120+X123+X125+X94+X100+X68+X55
        # formulaSelfComposed <- emotion ~ X109+X48+X54+X117+X103+X89+X99+X121+X127+X91+X53+X116+X135+X125+ #X122
        # mouseChange   <- emotion ~ X116+X117+X118+X119+X120+X121+X122+X123+X124+X125+X126+X127
        # +X48+X49+X50+X51+X52+X53+X54+X55+X56+X57+X58+X59
        # browsChange   <- emotion ~ X85+X86+X87+X88+X89+X90+X91+X92+X93+X94 + X17+X18+X19+X20+X21+X22+X23+X24+X25+X26
        # noseChange    <- emotion ~ X99+X100+X101+X102+X103 + X31+X32+X33+X34+X35
        # eyesChange    <- emotion ~ X36+X37+X38+X39+X40+X41+X42+X43+X44+X45+X46+X47
        # +X104+X105+X106+X107+X108+X109+X110+X111+X112+X113+X114+X115
        #+X48+X54+X116+X122
        # optimalChange2 <- emotion ~ X90+X35+X99+X17+X91+X104+X28+X80+X75+X82+X98+X83+X102+X107+X89+X8+X96+X38+X74+X17+X101+X77+X5
        # optimalCfs <- emotion ~ X26+X53+X71+X77+X78+X104+X105+X117+X124+X125+X126
        # withoutJaw <- emotion ~ X17+X18+X19+X20+X21+X22+X23+X24+X25+X26+X27+X28+X29+X30+X31+X32+X33+X34+X35+X36+X37+X38+X39+X40+X41+X42+X43+X44+X45+X46+X47+X48+X49+X50+X51+X52+X53+X54+X55+X56+X57+X58+X59+X60+X61+X62+X63+X64+X65+X66+X67+X68+X85+X86+X87+X88+X89+X90+X91+X92+X93+X94+X95+X96+X97+X98+X99+X100+X101+X102+X103+X104+X105+X106+X107+X108+X109+X110+X111+X112+X113+X114+X115+X116+X117+X118+X119+X120+X121+X122+X123+X124+X125+X126+X127+X128+X129+X130+X131+X132+X133+X134+X135+X136+X76
        # optimalForwardSearch <- emotion ~ X66+X76+X91+X117+X119+X121+X128 #+X136#+X112+X113
        # optBin <- emotion ~ X46+X57+X59+X87+X103 # X90+X100+X106+X121+X134 #X89+X106+X117+X135 #X19+X55+X128 X37+X49+X55+X57+X99+X104+X117+X121+X127 #X54+X103+X124 #X41+X54+X93+X119+X124+X127
        # optDif <- emotion ~ X23+X30+X57 #X23+X30+X57+X76 # X66+X91+X133 # X36+X48+X52+X134 #X30+X65+X103+X112+X122 # # Alt:               X35+X76+X101+X121+X122+X124+X128+X130
        optimalChange <- emotion ~ X52+X87+X89+X92+X103+X102+X101+X100+X117+X118+X120+X123+X125+X1+X23+X40+X24+X99+X38+X42+X45+X67+X88
        formula <- optimalChange #optBin # optimalForwardSearch #optimalForwardSearchGLOB #optimalChange
    }
    fit <- rpart(formula, method = "class", data = dataSet, control = rpart.control(minsplit = 1, cp = 0.006147541))
    fit
}

trainTestClassifier <- function(trainFolder = trainingDataFolder, testFolder = testDataFolder) {
    trainData  <- loadData(trainFolder)
    testData   <- loadData(testFolder)
    trueLabels <- testData[, 137]
    fit <- fitDTClassifier(trainData)
    res <- testDTClassifier(fit, testData, trueLabels)
    res / length(trueLabels)
}