Fix readme, fix MLJ sprinkle. Bump version

Project.toml

@@ -3,7 +3,7 @@ uuid = "e54bda2e-c571-11ec-9d64-0242ac120002"
 license = "MIT"
 desc = "Julia implementation of Modal Decision Trees and Random Forest algorithms"
 authors = ["Giovanni PAGLIARINI"]
-version = "0.1.2"
+version = "0.1.3"
 
 [deps]
 AbstractTrees = "1520ce14-60c1-5f80-bbc7-55ef81b5835c"

README.md

@@ -35,7 +35,7 @@ Additionally, these models:
 Simply type the following commands in Julia's REPL:
 
 ```julia
-using Pkg; Pkg.add(url="https://github.com/aclai-lab/ModalDecisionTrees.jl")
+using Pkg; Pkg.add("ModalDecisionTrees");
 ```
 -->
 
@@ -45,15 +45,16 @@ Simply type the following commands in Julia's REPL:
 
 ```julia
 # Install package
-using Pkg; Pkg.add(url="https://github.com/aclai-lab/ModalDecisionTrees.jl")
+using Pkg; Pkg.add("MLJ");
+using Pkg; Pkg.add("ModalDecisionTrees");
 
 # Import packages
 using MLJ
 using ModalDecisionTrees
 using Random
 
 # Load an example dataset (a temporal one)
-X, y = load_japanesevowels()
+X, y = ModalDecisionTrees.load_japanesevowels()
 N = length(y)
 
 # Instantiate an MLJ machine based on a Modal Decision Tree with ≥ 4 samples at leaf
@@ -67,8 +68,8 @@ train_idxs, test_idxs = p[1:round(Int, N*.8)], p[round(Int, N*.8)+1:end]
 fit!(mach, rows=train_idxs)
 
 # Perform predictions, compute accuracy
-yhat = predict(mach, X[test_idxs,:])
-accuracy = sum(yhat .== y[test_idxs])/length(yhat)
+yhat = predict_mode(mach, X[test_idxs,:])
+accuracy = MLJ.accuracy(yhat, y[test_idxs])
 
 # Print model
 report(mach).printmodel(3)

src/interfaces/MLJ.jl

@@ -118,7 +118,8 @@ function MMI.fit(m::SymbolicModel, verbosity::Integer, X, y, var_grouping, class
         printmodel                  = ModelPrinter(m, model, solemodel, var_grouping),
         sprinkle                    = (Xnew, ynew)->begin
             (Xnew, ynew, var_grouping, classes_seen, w) = MMI.reformat(m, Xnew, ynew; passive_mode = true)
-            translate_function(ModalDecisionTrees.sprinkle(model, Xnew, ynew))
+            preds, m = ModalDecisionTrees.sprinkle(model, Xnew, ynew)
+            preds, translate_function(m)
         end,
         # TODO remove redundancy?
         model                       = solemodel,

src/interfaces/MLJ/docstrings.jl

@@ -177,7 +177,7 @@ using Random
 tree = ModalDecisionTree(min_samples_leaf=4)
 
 # Load an example dataset (a temporal one)
-X, y = load_japanesevowels()
+X, y = ModalDecisionTrees.load_japanesevowels()
 N = length(y)
 
 mach = machine(tree, X, y)
@@ -190,8 +190,8 @@ train_idxs, test_idxs = p[1:round(Int, N*.8)], p[round(Int, N*.8)+1:end]
 fit!(mach, rows=train_idxs)
 
 # Perform predictions, compute accuracy
-yhat = predict(mach, X[test_idxs,:])
-accuracy = sum(yhat .== y[test_idxs])/length(yhat)
+yhat = predict_mode(mach, X[test_idxs,:])
+accuracy = MLJ.accuracy(yhat, y[test_idxs])
 
 # Access raw model
 fitted_params(mach).model
@@ -277,7 +277,7 @@ using Random
 forest = ModalRandomForest(ntrees = 50)
 
 # Load an example dataset (a temporal one)
-X, y = load_japanesevowels()
+X, y = ModalDecisionTrees.load_japanesevowels()
 N = length(y)
 
 mach = machine(forest, X, y)
@@ -291,9 +291,9 @@ fit!(mach, rows=train_idxs)
 
 # Perform predictions, compute accuracy
 Xnew = X[test_idxs,:]
-yhat = predict(mach, Xnew) # probabilistic predictions
-ynew = predict_mode(mach, Xnew)   # point predictions
-accuracy = sum(ynew .== y[test_idxs])/length(yhat)
+ynew = predict_mode(mach, Xnew) # point predictions
+accuracy = MLJ.accuracy(ynew, y[test_idxs])
+yhat = predict_mode(mach, Xnew) # probabilistic predictions
 pdf.(yhat, "1")    # probabilities for one of the classes ("1")
 
 # Access raw model

src/other/example-datasets.jl

@@ -38,7 +38,7 @@ function load_japanesevowels()
     minimum_n_points = minimum(collect(Iterators.flatten(eachrow(length.(X[:,Not([:speaker, :take, :utterance])])))))
     new_X = (x->x[1:minimum_n_points]).(X[:,Not([:speaker, :take, :utterance])])
 
-    new_X, varnames = SoleData.dataframe2cube(new_X)
+    # new_X, varnames = SoleData.dataframe2cube(new_X)
 
     new_X, Y
 end

test/classification/japanesevowels.jl

@@ -14,6 +14,8 @@ t = ModalDecisionTree(;
 # Load an example dataset (a temporal one)
 X, y = ModalDecisionTrees.load_japanesevowels()
 
+X, varnames = SoleData.dataframe2cube(X)
+
 p = randperm(Random.MersenneTwister(2), 100)
 X, y = X[:, :, p], y[p]
 
@@ -93,6 +95,8 @@ end
 
 X, y = ModalDecisionTrees.load_japanesevowels()
 
+X, varnames = SoleData.dataframe2cube(X)
+
 multilogiset, var_grouping = ModalDecisionTrees.wrapdataset(X, ModalDecisionTree(; min_samples_leaf = 1))
 
 # A Modal Decision Tree