XgboostModel(params={
'max_depth': 3,
'eval_metric': 'mlogloss'
},
nrIterations=3)
]
# DNNModel(iterations=400)]
nrCopies = 40
origLen = len(modelList)
for i in range(nrCopies):
for reference in modelList[:origLen]:
modelList.extend([deepcopy(reference)])
print(modelList)
#train adaboost
adaBoost = AdaBoost(modelList)
trainTestSplits = 4
(accuracy, F1, logLoss) = adaBoost.crossValidate(xTrain,
yTrain,
trainTestSplits,
0.2,
verbose=True)
print(
'avg acc: {:2.4f}, avg F1: {:2.4f}, avg. log loss: {:2.4f} based on {:d} tt splits'
.format(accuracy, F1, logLoss, trainTestSplits))
#(predLabels, predProb, classToIndex) = adaBoost.learn(xTrain, yTrain, verbose=True)
#predLabelsTest, predProbTest = adaBoost.predict(xTest)
# uncomment to write the predictions for the test set to a file
#writeTestSet(list(testSet.df['VisitNumber']), classToIndex, predProbTest, os.path.join(datafolder, 'testpred.csv'))
