# imports
import weka.core.converters.ConverterUtils.DataSource as DS
import weka.filters.Filter as Filter
import weka.filters.unsupervised.attribute.Remove as Remove
import os

# load data
data = DS.read(os.environ.get("MOOC_DATA") + os.sep + "iris.arff")

# remove class attribute
rem = Remove()
rem.setOptions(["-R", "last"])
rem.setInputFormat(data)
dataNew = Filter.useFilter(data, rem)

# output filtered dataset
print(dataNew)

# imports
import weka.core.converters.ConverterUtils.DataSource as DS
import weka.classifiers.trees.J48 as J48
import os

# load data
data = DS.read(os.environ.get("MOOC_DATA") + os.sep + "anneal.arff")
data.setClassIndex(data.numAttributes() - 1)

# configure classifier
cls = J48()
cls.setOptions(["-C", "0.3"])

# build classifier
cls.buildClassifier(data)

# output model
print(cls)

# imports
import weka.core.converters.ConverterUtils.DataSource as DS
import weka.classifiers.Evaluation as Evaluation
import weka.classifiers.trees.J48 as J48
import java.util.Random as Random
import os

# load data
data = DS.read(os.environ.get("MOOC_DATA") + os.sep + "anneal.arff")
data.setClassIndex(data.numAttributes() - 1)

# configure classifier
cls = J48()
cls.setOptions(["-C", "0.3"])

# cross-validate classifier
evl = Evaluation(data)
evl.crossValidateModel(cls, data, 10, Random(1))

# print statistics
print(evl.toSummaryString("=== J48 on anneal (stats) ===", False))
print(evl.toMatrixString("=== J48 on anneal (confusion matrix) ==="))

# imports
import weka.classifiers.trees.J48 as J48
import weka.core.converters.ConverterUtils.DataSource as DS
import os

# load training data
data = DS.read(os.environ.get("MOOC_DATA") + os.sep + "anneal_train.arff")
data.setClassIndex(data.numAttributes() - 1)

# configure classifier
cls = J48()
cls.setOptions(["-C", "0.3"])

# build classifier on training data
cls.buildClassifier(data)

# load unlabeled data
dataUnl = DS.read(os.environ.get("MOOC_DATA") + os.sep + "anneal_unlbl.arff")
dataUnl.setClassIndex(dataUnl.numAttributes() - 1)

# test compatibility of train/unlabeled datasets
msg = dataUnl.equalHeadersMsg(data)
if msg is not None:
    print("train and prediction data are not compatible:\n" + msg)

# make predictions
for inst in dataUnl:
    dist = cls.distributionForInstance(inst)
    labelIndex = cls.classifyInstance(inst)
    label = dataUnl.classAttribute().value(int(labelIndex))
    print(str(dist) + " - " + str(labelIndex) + " - " + label)

# Note: install jfreechartOffscreenRenderer package as well for JFreeChart library

# imports
import weka.classifiers.Evaluation as Evaluation
import weka.classifiers.functions.LinearRegression as LinearRegression
import weka.core.converters.ConverterUtils.DataSource as DS
import java.util.Random as Random
import org.jfree.data.xy.DefaultXYZDataset as DefaultXYZDataset
import org.jfree.chart.ChartFactory as ChartFactory
import org.jfree.chart.plot.PlotOrientation as PlotOrientation
import org.jfree.chart.ChartPanel as ChartPanel
import org.jfree.chart.renderer.xy.XYBubbleRenderer as XYBubbleRenderer
import org.jfree.chart.ChartUtilities as ChartUtilities
import javax.swing.JFrame as JFrame
import java.io.File as File
import os

# load data
data = DS.read(os.environ.get("MOOC_DATA") + os.sep + "bodyfat.arff")
data.setClassIndex(data.numAttributes() - 1)

# configure classifier
cls = LinearRegression()
cls.setOptions(["-C", "-S", "1"])

# cross-validate classifier
evl = Evaluation(data)
evl.crossValidateModel(cls, data, 10, Random(1))

# collect predictions
act = []
prd = []
err = []
for i in range(evl.predictions().size()):
    prediction = evl.predictions().get(i)
    act.append(prediction.actual())
    prd.append(prediction.predicted())
    err.append(abs(prediction.actual() - prediction.predicted()))
    
# create plot
plotdata = DefaultXYZDataset()
plotdata.addSeries("LR on " + data.relationName(), [act, prd, err])
plot = ChartFactory.createScatterPlot(\
    "Classifier errors", "Actual", "Predicted", \
    plotdata, PlotOrientation.VERTICAL, True, True, True)
plot.getPlot().setRenderer(XYBubbleRenderer())

# display plot
frame = JFrame()
frame.setTitle("Weka")
frame.setSize(800, 800)
frame.setLocationRelativeTo(None)
frame.getContentPane().add(ChartPanel(plot))
frame.setVisible(True)

# imports
import weka.classifiers.bayes.BayesNet as BayesNet
import weka.core.converters.ConverterUtils.DataSource as DS
import weka.gui.graphvisualizer.GraphVisualizer as GraphVisualizer
import javax.swing.JFrame as JFrame
import os

# load data
data = DS.read(os.environ.get("MOOC_DATA") + os.sep + "iris.arff")
data.setClassIndex(data.numAttributes() - 1)

# configure classifier
cls = BayesNet()
cls.setOptions(["-Q", "weka.classifiers.bayes.net.search.local.K2", "--", "-P", "2"])

# build classifier
cls.buildClassifier(data)

# display tree
gv = GraphVisualizer()
gv.readBIF(cls.graph())
frame = JFrame("BayesNet - " + data.relationName())
frame.setDefaultCloseOperation(JFrame.DISPOSE_ON_CLOSE)
frame.setSize(800, 600)
frame.getContentPane().add(gv)
frame.setVisible(True)
    
# adjust tree layout
gv.layoutGraph()