def splitDataFirstMode():
# use all available data
sampleData = dp.returnData(1, randomSeed=10)
saveDataset(sampleData, 'AD_dataset')
# split data into X and y
da = DataAnalysis(dirName=dirName)
X, y = da.splitXY(sampleData)
# split data into training (80%) and testing (20%) set
xTrain, xTest, yTrain, yTest = da.splitTrainTest(X, y, trainSize=0.8)
saveDataset(xTrain.assign(normality=yTrain.values), 'AD_set_train')
saveDataset(xTest.assign(normality=yTest.values), 'AD_set_test')
# get data characteristics for current dataset size
dataInfo[1] = da.getDataCharacteristics(yTrain, yTest)
# Run predictions
for datasetSize in (selectedSizes or [0.2, 0.4, 0.6, 0.8, 1]):
# get no. of samples and prediction accuracy
noOfSamples, predictions = da.getScores(
xTrain,
xTest,
yTrain,
yTest,
trainSize=datasetSize,
randomSeeds=randomSeeds,
selectedAlgorithms=selectedAlgorithms
) if selectedAlgorithms else da.getScores(xTrain,
yTrain,
xTest,
yTest,
trainSize=datasetSize,
randomSeeds=randomSeeds)
# save results for graphs and .csv files
savePredictionScores(noOfSamples, predictions, datasetSize)
def normalMode():
da = DataAnalysis(dirName=dirName)
# Run predictions
for datasetSize in (selectedSizes or [0.2, 0.4, 0.6, 0.8, 1]):
# get the percentage of all data
sampleData = dp.returnData(datasetSize, randomSeed=10)
saveDataset(sampleData, 'AD_dataset')
# split data into X and y
X, y = da.splitXY(sampleData)
# split data into training (80%) and testing (20%) set
xTrain, xTest, yTrain, yTest = da.splitTrainTest(X, y, trainSize=0.8)
# get data characteristics for current dataset size
dataInfo[datasetSize] = da.getDataCharacteristics(yTrain, yTest)
# get no. of samples and prediction accuracy (trainSize is set to 1, so no further data splitting is done)
# multiple runs make no sense here, since we always take whole set
noOfSamples, predictions = da.getScores(
xTrain,
xTest,
yTrain,
yTest,
trainSize=1,
randomSeeds=[10],
selectedAlgorithms=selectedAlgorithms
) if selectedAlgorithms else da.getScores(xTrain,
yTrain,
xTest,
yTest,
trainSize=datasetSize,
randomSeeds=randomSeeds)
# save results for graphs and .csv files
savePredictionScores(noOfSamples, predictions, datasetSize)
