diff --git a/src/Maxnet.jl b/src/Maxnet.jl
index 776648e..f01db6c 100644
--- a/src/Maxnet.jl
+++ b/src/Maxnet.jl
@@ -11,6 +11,7 @@ export IdentityLink, CloglogLink, LogitLink, LogLink # re-export relevant links
 export LassoBackend, GLMNetBackend
 export maxnet, predict
 export LinearFeature, CategoricalFeature, QuadraticFeature, ProductFeature, ThresholdFeature, HingeFeature
+export MaxnetBinaryClassifier
 
 # Write your package code here.
 
diff --git a/src/feature_classes.jl b/src/feature_classes.jl
index b23a97c..e639b94 100644
--- a/src/feature_classes.jl
+++ b/src/feature_classes.jl
@@ -26,6 +26,17 @@ function features_from_string(s::AbstractString)
 end
 
 # Default features based on number of presences
+
+"""
+    default_features(np)
+
+Takes the number of presences `np` and returns a `Vector` of `AbstractFeatureClass`s that are used my maxent as default.
+
+If `np` is less than ten, then only `LinearFeature` and `CategoricalFeature` are used.
+If it is at least 10, then `QuadraticFeature` is additionally used.
+If it is at least 15, then `HingeFeature` is additionally used.
+If it is at least 80, then `ProductFeature` is additionally used.
+"""
 function default_features(np)
     features = [LinearFeature(), CategoricalFeature()]
     if np >= 10
diff --git a/src/maxnet_function.jl b/src/maxnet_function.jl
index 1682c08..e0ddf26 100644
--- a/src/maxnet_function.jl
+++ b/src/maxnet_function.jl
@@ -23,7 +23,8 @@ end
 
 """
     maxnet(
-        presences, predictors, [features];
+        presences, predictors; 
+        [features],
         regularization_multiplier, regularization_function,
         addsamplestobackground, weight_factor, backend, 
         kw...
@@ -36,20 +37,28 @@ end
 # Keywords
 - `features`: Either:
     - A `Vector` of `AbstractFeatureClass` type features; or
-    - A string where "l" = linear and categorical, "q" = quadratic, "p" = product, "t" = threshold, "h" = hinge; or
-    - Nothing, in which case the default features based on the number of presences are used
+    - A `String` where "l" = linear and categorical, "q" = quadratic, "p" = product, "t" = threshold, "h" = hinge (e.g. "lqh"); or
+    - The default, in which case the features are based on the number of presences are used. See [`default_features`](@ref)
 - `regularization_multiplier`: A constant to adjust regularization, where a higher `regularization_multiplier` results in a higher penalization for features
 - `regularization_function`: A function to compute a regularization for each feature. A default `regularization_function` is built in.
 - `addsamplestobackground`: A boolean, where `true` adds the background samples to the predictors. Defaults to `true`.
 - `n_knots`: the number of knots used for Threshold and Hinge features. Defaults to 50. Ignored if there are neither Threshold nor Hinge features
-- `weight_factor`: A `Float64` to adjust the weight of the background samples. Defaults to 100.0.
-- `backend`: Either `LassoBackend()` or `GLMNetBackend()`, to use Lasso.jl or GLMNet.jl fit the model.
+- `weight_factor`: A `Float64` value to adjust the weight of the background samples. Defaults to 100.0.
+- `backend`: Either `LassoBackend()` or `GLMNetBackend()`, to use either Lasso.jl or GLMNet.jl to fit the model.
 Lasso.jl is written in pure julia, but can be slower with large model matrices (e.g. when hinge is enabled). Defaults to `LassoBackend`.
-- `kw...`: Further arguments to be passed to Lasso.fit or GLMNet.glmnet
+- `kw...`: Further arguments to be passed to `Lasso.fit` or `GLMNet.glmnet`
 
 # Returns
 - `model`: A model of type `MaxnetModel`
 
+# Examples
+```jldoctest
+    using Maxnet
+    p_a, env = Maxnet.bradypus()
+
+    bradypus_model = maxnet(p_a, env; features = "lq", backend = GLMNetBackend());
+```	
+
 """
 function maxnet(
     presences::BitVector, predictors; 
diff --git a/src/mlj_interface.jl b/src/mlj_interface.jl
index e04d302..0ead673 100644
--- a/src/mlj_interface.jl
+++ b/src/mlj_interface.jl
@@ -1,12 +1,3 @@
-#=MMI.@mlj_model mutable struct MaxnetBinaryClassifier <: MMI.Deterministic
-    features = ""
-    regularization_multiplier = 1.0
-    regularization_function = default_regularization
-    weight_factor = 100.
-    backend = LassoBackend
-    kw...
-end
-=#
 mutable struct MaxnetBinaryClassifier <: MMI.Probabilistic
     features::Union{String, Vector{<:AbstractFeatureClass}}
     regularization_multiplier::Float64
@@ -14,6 +5,7 @@ mutable struct MaxnetBinaryClassifier <: MMI.Probabilistic
     weight_factor::Float64
     backend::MaxnetBackend
     link::GLM.Link
+    clamp::Bool
     kw
 end
 
@@ -21,15 +13,39 @@ function MaxnetBinaryClassifier(;
     features="", 
     regularization_multiplier = 1.0, regularization_function = default_regularization, 
     weight_factor = 100., backend = LassoBackend(), 
-    link = CloglogLink(),
+    link = CloglogLink(), clamp = false,
     kw...
 )
+
     MaxnetBinaryClassifier(
         features, regularization_multiplier, regularization_function, 
-        weight_factor, backend, link, kw
+        weight_factor, backend, link, clamp, kw
     )
 end
 
+"""
+    MaxnetBinaryClassifier
+
+    A model type for fitting a maxnet model using `MLJ`.
+        
+    Use `MaxnetBinaryClassifier()` to create an instance with default parameters, or use keyword arguments to specify parameters.
+    
+    The `link` and `clamp` keywords are passed to `predict`. All other keywords are passed to `maxnet` when te model is fit.
+    See the documentation of [`maxnet`](@ref) for the parameters and their defaults.
+
+    # Examples
+    ```jldoctest
+    using Maxnet, MLJBase
+    p_a, env = Maxnet.bradypus()
+
+    mach = machine(MaxnetBinaryClassifier(features = "lqp"), env, categorical(p_a))
+    fit!(mach)
+    yhat = MLJBase.predict(mach, env)
+    ```
+
+"""
+MaxnetBinaryClassifier
+
 MMI.input_scitype(::Type{<:MaxnetBinaryClassifier}) =
         MMI.Table{<:Union{<:AbstractVector{<:Continuous}, <:AbstractVector{<:Multiclass}}} #{<:Union{<:Continuous <:Multiclass}}
     
@@ -65,6 +81,6 @@ function MMI.fit(m::MaxnetBinaryClassifier, verbosity::Int, X, y)
 end
 
 function MMI.predict(m::MaxnetBinaryClassifier, (fitresult, decode), Xnew)
-    p = predict(fitresult, Xnew; link = m.link)
+    p = predict(fitresult, Xnew; link = m.link, clamp = m.clamp)
     MMI.UnivariateFinite(decode, [1 .- p, p])
 end
\ No newline at end of file
diff --git a/src/predict.jl b/src/predict.jl
index ca01e39..f92a7ae 100644
--- a/src/predict.jl
+++ b/src/predict.jl
@@ -1,11 +1,9 @@
-""""
+"""
     predict(m, x; link, clamp)
 
-    Use a fit maxnet model to predict
-
 # Arguments
-- `m`: a MaxnetModel generated by maxnet()
-- `x`: a Tables.jl-compatible table of predictors. All columns that were used to fit `m` should be present in `x`
+- `m`: a MaxnetModel as returned by `maxnet`
+- `x`: a `Tables.jl`-compatible table of predictors. All columns that were used to fit `m` should be present in `x`
 
 # Keywords
 - `link`: the link function used. Defaults to CloglogLink(), which is the default on the Maxent Java appliation since version 4.3.
@@ -15,10 +13,9 @@
 - `clamp`: If `true`, values in `x` will be clamped to the range the model was trained on. Defaults to `false`.
 
 # Returns
-A `Vector` of prediction values.
+A `Vector` with the resulting predictions.
 
 """
-
 function predict(m::MaxnetModel, x; link = CloglogLink(), clamp = false)
     predictors = Tables.columntable(x)
     for k in keys(predictors)