Dev

README.md

@@ -87,24 +87,24 @@ config = EvoTreeRegressor(
     min_weight=1.0,
     rowsample=0.5, 
     colsample=1.0)
-model = fit_evotree(config; x_train = x_train, y_train = y_train)
-preds = predict(model, x_eval)
+m = fit_evotree(config; x_train, y_train)
+preds = m(x_train)
 ```
 
 ## Feature importance
 
 Returns the normalized gain by feature.
 
 ```julia
-features_gain = importance(model)
+features_gain = importance(m)
 ```
 
 ## Plot
 
 Plot a given tree of the model:
 
 ```julia
-plot(model, 2)
+plot(m, 2)
 ```
 
 ![](figures/plot_tree.png)
@@ -114,6 +114,8 @@ Note that 1st tree is used to set the bias so the first real tree is #2.
 ## Save/Load
 
 ```julia
-EvoTrees.save(model, "data/model.bson")
-model = EvoTrees.load("data/model.bson");
-```
+EvoTrees.save(m, "data/model.bson")
+m = EvoTrees.load("data/model.bson");
+```
+
+A GPU model should be converted into a CPU one before saving: `m_cpu = convert(EvoTree, m_gpu)`.

docs/src/index.md

@@ -21,3 +21,13 @@ From General Registry:
 ```julia-repl
 julia> Pkg.add("EvoTrees")
 ```
+
+
+## Save/Load
+
+```julia
+EvoTrees.save(m, "data/model.bson")
+m = EvoTrees.load("data/model.bson");
+```
+
+A GPU model should be converted into a CPU one before saving: `m_cpu = convert(EvoTree, m_gpu)`.

experiments/MLJ.jl

@@ -8,7 +8,6 @@ using CategoricalArrays
 using Distributions
 using EvoTrees
 using EvoTrees: logit, sigmoid
-# import EvoTrees: EvoTreeRegressor, EvoTreeClassifier, EvoTreeCount, EvoTreeGaussian
 
 ##################################################
 ### Regression - small data
@@ -74,7 +73,7 @@ config = EvoTreeClassifier(
 model = fit_evotree(config; x_train, y_train);
 model = fit_evotree(config; x_train, y_train, x_eval = x_train, y_eval = y_train, metric=:mlogloss, print_every_n=10, early_stopping_rounds=25);
 
-pred = predict(model, x_train)
+pred = model(x_train)
 pred_cat = pred .> 0.5
 sum((y_train .== "B") .== pred_cat[:, 1]) / length(y_train)
 
@@ -90,16 +89,16 @@ fit!(mach, rows = train, verbosity = 1)
 rpt = report(mach)
 MLJBase.feature_importances(config, mach.fitresult, rpt)
 
-pred_train = predict(tree, selectrows(X, train))
-pred_train_mode = predict_mode(tree, selectrows(X, train))
-println(cross_entropy(pred_train, selectrows(y, train)) |> mean)
-println(sum(pred_train_mode .== y[train]))
+pred_train = EvoTrees.predict(mach, selectrows(X, train))
+pred_train_mode = predict_mode(mach, selectrows(X, train))
+println(cross_entropy(pred_train, selectrows(y_train, train)) |> mean)
+println(sum(pred_train_mode .== y_train[train]) / length(train))
 
-pred_test = predict(tree, selectrows(X, test))
-pred_test_mode = predict_mode(tree, selectrows(X, test))
-println(cross_entropy(pred_test, selectrows(y, test)) |> mean)
-println(sum(pred_test_mode .== y[test]))
-pred_test_mode = predict_mode(tree, selectrows(X, test))
+pred_test = EvoTrees.predict(mach, selectrows(X, test))
+pred_test_mode = predict_mode(mach, selectrows(X, test))
+println(cross_entropy(pred_test, selectrows(y_train, test)) |> mean)
+println(sum(pred_test_mode .== y_train[test]) / length(test))
+pred_test_mode = predict_mode(mach, selectrows(X, test))
 
 # using LossFunctions, Plots
 # evo_model = EvoTreeClassifier(max_depth=6, η=0.05, λ=1.0, γ=0.0, nrounds=10, nbins=64)

experiments/benchmarks-softmax.jl

@@ -39,11 +39,10 @@ params_evo = EvoTreeClassifier(
     gamma=0.0,
     eta=0.05,
     max_depth=6,
-    min_weight=10.0,
+    min_weight=1.0,
     rowsample=0.5,
     colsample=0.5,
-    nbins=64,
-    num_class=num_class
+    nbins=64
 )
 
 nobs = Int(1e6)
@@ -65,7 +64,7 @@ params_evo.device = "cpu"
 @time m_evo = fit_evotree(params_evo; x_train, y_train, x_eval=x_train, y_eval=y_train, metric=metric_evo, print_every_n=100);
 @time m_evo = fit_evotree(params_evo; x_train, y_train, x_eval=x_train, y_eval=y_train, metric=metric_evo, print_every_n=100);
 # @btime fit_evotree($params_evo; x_train=$x_train, y_train=$y_train, x_eval=$x_train, y_eval=$y_train, metric=metric_evo);
-# @btime fit_evotree($params_evo; x_train=$x_train, y_train=$y_train);
+@time fit_evotree(params_evo; x_train, y_train);
 @info "evotrees predict CPU:"
 @time pred_evo = EvoTrees.predict(m_evo, x_train);
 @btime EvoTrees.predict($m_evo, $x_train);

experiments/readme_plots_cpu.jl

@@ -38,7 +38,6 @@ params1 = EvoTreeRegressor(T=Float32,
 @time model = fit_evotree(params1; x_train, y_train, x_eval, y_eval, metric=:mse, print_every_n=25, early_stopping_rounds=20);
 # 67.159 ms (77252 allocations: 28.06 MiB)
 # @time model = fit_evotree(params1, X_train, Y_train, X_eval = X_eval, Y_eval = Y_eval, print_every_n = 999);
-# @btime model = fit_evotree($params1, $X_train, $Y_train, X_eval = $X_eval, Y_eval = $Y_eval);
 # Profile.clear()  # in case we have any previous profiling data
 # @profile fit_evotree(params1, X_train, Y_train, X_eval = X_eval, Y_eval = Y_eval, print_every_n = 25)
 # ProfileView.view()