[doc] scikit-learn and h2o (#64)

* update the scikitlearn example

* modelStudio alias for dalex Explainer

* add h2o example

Author: hbaniecki

NAMESPACE

@@ -1,6 +1,8 @@
 # Generated by roxygen2: do not edit by hand
 
+S3method(modelStudio,dalex._explainer.object.Explainer)
 S3method(modelStudio,explainer)
+S3method(modelStudio,python.builtin.object)
 export(modelStudio)
 export(modelStudioOptions)
 import(progress)

R/modelStudio.R

@@ -384,6 +384,15 @@ modelStudio.explainer <- function(explainer,
   model_studio
 }
 
+#:# alias for reticulate pickle/dalex Explainer
+#' @noRd
+#' @export
+modelStudio.python.builtin.object <- modelStudio.explainer
+
+#' @noRd
+#' @export
+modelStudio.dalex._explainer.object.Explainer <- modelStudio.explainer
+
 #' @noRd
 #' @title remove_file_paths
 #'

README.md

@@ -16,7 +16,7 @@ The main `modelStudio()` function computes various (instance and dataset level)
 [**explain FIFA20**](https://pbiecek.github.io/explainFIFA20/)  
 [explain Lung Cancer](https://github.com/hbaniecki/transparent_xai/)  
 [**R & Python examples**](http://modelstudio.drwhy.ai/articles/vignette_examples.html)  
-[More Resources](https://modeloriented.github.io/modelStudio/#more)  
+[More Resources](http://modelstudio.drwhy.ai/#more-resources)  
 [**FAQ & Troubleshooting**](https://github.com/ModelOriented/modelStudio/issues/54)
 
 ![](man/figures/short.gif)
@@ -73,7 +73,7 @@ install.packages("iBreakDown")
 
 # packages for explainer objects
 install.packages("DALEX")
-devtools::install_github("ModelOriented/DALEXtra")
+install.packages("DALEXtra")
 ```
 
 ### mlr [dashboard](https://modeloriented.github.io/modelStudio/mlr.html)
@@ -94,7 +94,7 @@ test <- data[-index, ]
 # mlr ClassifTask takes target as factor
 train$survived <- as.factor(train$survived)
 
-# prepare the model
+# fit a model
 task <- makeClassifTask(id = "titanic",
                         data = train,
                         target = "survived")
@@ -137,7 +137,7 @@ test <- data[-index, ]
 train_matrix <- model.matrix(survived ~.-1, train)
 test_matrix <- model.matrix(survived ~.-1, test)
 
-# prepare the model
+# fit a model
 xgb_matrix <- xgb.DMatrix(train_matrix, label = train$survived)
 params <- list(eta = 0.01, subsample = 0.6, max_depth = 7, min_child_weight = 3,
                objective = "binary:logistic", eval_metric = "auc")
@@ -161,96 +161,88 @@ modelStudio(explainer,
 
 ### scikit-learn [dashboard](https://modeloriented.github.io/modelStudio/scikit-learn.html)
 
-Use `pickle` Python module and `reticulate` R package to easily produce modelStudio for scikit-learn model.
+Use `pickle` Python module and `reticulate` R package to easily make a studio for a scikit-learn model.
 
-In this example we fit a Pipeline MLPClassifier on the titanic data. First install the `dalex` package.
+In this example we will fit a Pipeline MLPClassifier model on titanic data. 
+
+Install the `dalex` package.
 
 ```bash
 pip3 install dalex --force
 ```
 
-Make an explainer object in Python:
+First, use `dalex` in Python:
 
 ```python
-# import modules
+# load packages and data
 import dalex as dx
-from dalex import datasets
 
-from sklearn.neural_network import MLPClassifier
-from sklearn.preprocessing import StandardScaler, LabelEncoder, OneHotEncoder
-from sklearn.impute import SimpleImputer
+from sklearn.model_selection import train_test_split
 from sklearn.pipeline import Pipeline
-from sklearn.tree import DecisionTreeRegressor
+from sklearn.preprocessing import StandardScaler, OneHotEncoder
+from sklearn.impute import SimpleImputer
 from sklearn.compose import ColumnTransformer
+from sklearn.neural_network import MLPClassifier
 
-# load the data
-data = datasets.load_titanic()
+data = dx.datasets.load_titanic()
 X = data.drop(columns='survived')
 y = data.survived
 
-# make a pipeline model
+# split the data
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=1)
+
+# fit a pipeline model
 numeric_features = ['age', 'fare', 'sibsp', 'parch']
-numeric_transformer = Pipeline(steps=[
+numeric_transformer = Pipeline(
+  steps=[
     ('imputer', SimpleImputer(strategy='median')),
-    ('scaler', StandardScaler())])
-
+    ('scaler', StandardScaler())
+    ]
+)
 categorical_features = ['gender', 'class', 'embarked']
-categorical_transformer = Pipeline(steps=[
+categorical_transformer = Pipeline(
+  steps=[
     ('imputer', SimpleImputer(strategy='constant', fill_value='missing')),
-    ('onehot', OneHotEncoder(handle_unknown='ignore'))])
+    ('onehot', OneHotEncoder(handle_unknown='ignore'))
+    ]
+)
 
 preprocessor = ColumnTransformer(
-    transformers=[
-        ('num', numeric_transformer, numeric_features),
-        ('cat', categorical_transformer, categorical_features)])
-
-
-clf = Pipeline(steps=[('preprocessor', preprocessor),
-                      ('classifier', MLPClassifier(hidden_layer_sizes=(150,100,50),
-                                                   max_iter=500, random_state=0))])
-
-clf.fit(X, y)
+  transformers=[
+    ('num', numeric_transformer, numeric_features),
+    ('cat', categorical_transformer, categorical_features)
+    ]
+)
+
+model = Pipeline(
+  steps=[
+    ('preprocessor', preprocessor),
+    ('classifier', MLPClassifier(hidden_layer_sizes=(150,100,50), max_iter=500, random_state=0))
+    ]
+)
+model.fit(X_train, y_train)
 
-# make an explainer
-explainer = dx.Explainer(clf, X, y)
+# create an explainer for the model
+explainer = dx.Explainer(model, X_test, y_test, label = 'scikit-learn')
 
-# remove these functions before dump
+#! remove residual_function before dump !
 explainer.residual_function = None
-explainer.predict_function = None
 
 # pack the explainer into a pickle file
-import pickle        
-pickle_out = open("explainer_titanic.pickle","wb")
+import pickle
+pickle_out = open("explainer_scikitlearn.pickle","wb")
 pickle.dump(explainer, pickle_out)
-pickle_out.close()     
+pickle_out.close()
 ```
 
-Then use `modelStudio` in R:
+Then, use `modelStudio` in R:
 
 ```r
-# use reticulate to load the explainer from a pickle file
+# load the explainer from the pickle file
 library(reticulate)
-explainer <- py_load_object('explainer_titanic.pickle')
-
-# make a predict_function
-predict_function <- function(model, data) {
-  if ("predict_proba" %in% names(model)) {
-    pred <- model$predict_proba(data)
-    if (ncol(pred) == 2) {
-      pred <- pred[,2]
-    }
-  } else {
-    pred <- model$predict(data)
-  }
-  pred
-}
-
-# adjust the explainer
-explainer$predict_function <- predict_function
-explainer$label <- 'scikit-learn'
-class(explainer) <- c(class(explainer), 'explainer')
-
-# make a modelStudio
+explainer <- py_load_object('explainer_scikitlearn.pickle', pickle = "pickle")
+
+# make a studio for the model
 library(modelStudio)
 modelStudio(explainer)
 ```

inst/WORDLIST

@@ -28,6 +28,7 @@ tensorflow
 Shapley
 cran
 CRAN
-MLPCLassifier
+MLPClassifier
 keras
 lightGBM
+customizable

pkgdown/_pkgdown.yml

@@ -3,4 +3,5 @@ template:
   default_assets: false
   params:
     ganalytics: UA-5650686-14
-    noindex: true
+    noindex: true
+