#HousingTrain,HousingTest = InitializeTestSet.SplitTrainTest(housing, 0.20)
HousingTrain, HousingTest = InitializeTestSet.IncomeStratifiedSplitting(
housing)
Visualization.LongLatPlot(
housing, 'GeoPlot.pdf', False
) #This is a coloured plot of the housing data in a longitude-latitude axes
CorrMatrix = housing.corr(
) #This is the correlation matrix for the housing data set
Visualization.PlotCorrMatrix(housing, 'CorrelationMatrix.pdf', False)
Visualization.PlotMedianIncomeCorrMatrix(housing, 'MedianIncomeScatter.pdf',
False)
#Create new attributes
housing["rooms_per_household"] = housing["total_rooms"] / housing["households"]
housing[
"bedrooms_per_room"] = housing["total_bedrooms"] / housing["total_rooms"]
housing[
"population_per_household"] = housing["population"] / housing["households"]
housing2 = HousingTrain.drop(
"median_house_value",
axis=1) #Creates a copy of the data in "median housing value"
hosuing_labels = HousingTrain["median_house_value"].copy()
housing_tr = Preprocessing.CleanDataHoles(
