Visualize the Probability Distributions of Classifiers 

Generate clustered data by sampling normal distributions.

In[1]:=

X sampledata[center_] := RandomVariate[MultinormalDistribution[center, IdentityMatrix[2]], 30]; clusters = sampledata /@ {{2, 2}, {-2, 2}, {0, -3}, {0, -0}}; colors = {Yellow, Blue, Green, Red}; ListPlot[clusters, PlotStyle -> Darker@colors]

Out[1]=

Visualize class probabilities as a function of the features for various classifiers.

show complete Wolfram Language inputhide input

In[2]:=

X plotprobabilities[method_] := Module[{c, proba}, c = Classify[<|Thread[colors -> clusters]|>, Method -> method]; proba = Table[ Append[{x, y}, #] & /@ Lookup[c[{x, y}, "Probabilities"], colors], {x, -4, 4, .5}, {y, -5, 4, .5}]; proba = Flatten[#, 1] & /@ Transpose[proba, {3, 2, 1, 4}]; Show[ ListPlot3D[proba, PlotRange -> {0, 1}, PlotLabel -> method], ListPointPlot3D[Map[Append[#, 1] &, clusters, {2}], PlotStyle -> ({Opacity[0.5], #} & /@ colors)] ] ]; plotprobabilities /@ {"LogisticRegression", "NearestNeighbors", "RandomForest", "SupportVectorMachine", "NaiveBayes", "NeuralNetwork"}

Out[2]=