✨ Add BalancedModel

.github/workflows/CI.yml

@@ -19,11 +19,8 @@ jobs:
       fail-fast: false
       matrix:
         version:
-          - '1.0'
           - '1.8'
-          - 'nightly'
-        os:
-          - ubuntu-latest
+        os: [ubuntu-latest, windows-latest, macOS-latest]
         arch:
           - x64
     steps:

.gitignore

@@ -1 +1,2 @@
 /Manifest.toml
+.CondaPkg

Project.toml

@@ -10,17 +10,16 @@ OrderedCollections = "bac558e1-5e72-5ebc-8fee-abe8a469f55d"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 
 [compat]
-julia = "1"
 MLJBase = "0.21"
-MLJModelInterface = "1.9"
 OrderedCollections = "1.6"
+julia = "1.6"
 
 [extras]
 DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
 Imbalance = "c709b415-507b-45b7-9a3d-1767c89fde68"
-MLJLIBSVMInterface = "61c7150f-6c77-4bb1-949c-13197eac2a52"
 MLJLinearModels = "6ee0df7b-362f-4a72-a706-9e79364fb692"
+MLJModels = "d491faf4-2d78-11e9-2867-c94bc002c0b7"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["Test", "Imbalance", "DataFrames", "MLJLIBSVMInterface", "MLJLinearModels"]
+test = ["Test", "Imbalance", "DataFrames", "MLJLinearModels", "MLJModels"]

README.md

@@ -1,2 +1,56 @@
 # MLJBalancing
-A package with exported learning networks that combine resampling methods from Imbalance.jl and classification models from MLJ
+A package providing composite models wrapping class imbalance algorithms from [Imbalance.jl](https://github.com/JuliaAI/Imbalance.jl).
+
+## ⏬ Instalattion
+```julia
+import Pkg;
+Pkg.add("MLJBalancing")
+```
+
+## 🚅 Sequential Resampling
+
+This package allows chaining of resampling methods from Imbalance.jl with classification models from MLJ. Simply construct a `BalancedModel` object while specifying the model (classifier) and an arbitrary number of resamplers (also called *balancers* - typically oversamplers and/or under samplers).
+
+### 📖 Example
+
+#### Construct the resamplers and the model
+```julia
+SMOTENC = @load SMOTENC pkg=Imbalance verbosity=0
+TomekUndersampler = @load TomekUndersampler pkg=Imbalance verbosity=0
+
+oversampler = SMOTENC(k=5, ratios=1.0, rng=42)
+undersampler = TomekUndersampler(min_ratios=0.5, rng=42)
+
+logistic_model = LogisticClassifier()
+```
+
+#### Wrap them all in BalancedModel
+```julia
+balanced_model = BalancedModel(model=logistic_model, balancer1=oversampler, balancer2=undersampler)
+```
+Here training data will be passed to `balancer1` then `balancer2`, whose output is used to train the classifier `model`.  In prediction, the resamplers `balancer1` and `blancer2` are bypassed. 
+
+In general, there can be any number of balancers, and the user can give the balancers arbitrary names. 
+
+#### At this point, they behave like one single model
+You can fit, predict, cross-validate and finetune it like any other MLJ model. Here is an example for finetuning
+```julia
+r1 = range(balanced_model, :(balancer1.k), lower=3, upper=10)
+r2 = range(balanced_model, :(balancer2.min_ratios), lower=0.1, upper=0.9)
+
+tuned_balanced_model = TunedModel(
+    model=balanced_model,
+    tuning=Grid(goal=4),
+    resampling=CV(nfolds=4),
+    range=[r1, r2],
+    measure=cross_entropy
+);
+
+mach = machine(tuned_balanced_model, X, y);
+fit!(mach, verbosity=0);
+fitted_params(mach).best_model
+```
+
+## 🚆🚆 Parallel Resampling with EasyEnsemble
+
+Coming soon...