Python sklearn.tree() Examples

def _get_tree_paths(tree, node_id, depth=0):
    """
    Returns all paths through the tree as list of node_ids
    """
    if node_id == _tree.TREE_LEAF:
        raise ValueError("Invalid node_id %s" % _tree.TREE_LEAF)

    left_child = tree.children_left[node_id]
    right_child = tree.children_right[node_id]

    if left_child != _tree.TREE_LEAF:
        left_paths = _get_tree_paths(tree, left_child, depth=depth + 1)
        right_paths = _get_tree_paths(tree, right_child, depth=depth + 1)

        for path in left_paths:
            path.append(node_id)
        for path in right_paths:
            path.append(node_id)
        paths = left_paths + right_paths
    else:
        paths = [[node_id]]
    return paths 

def visualize_tree(tree, feature_names, save_dir='./'):
    """Create tree png using graphviz.

    Args
    ----
    tree -- scikit-learn DecsisionTree.
    feature_names -- list of feature names.
    """
    with open(save_dir+'/'+"dt.dot", 'w') as f:
        export_graphviz(tree, out_file=f,
                        feature_names=feature_names)

    command = ["dot", "-Tpng", save_dir+"/dt.dot", "-o", save_dir+"/dt.png"]
    try:
        subprocess.check_call(command)
    except:
        exit("Could not run dot, ie graphviz, to "
             "produce visualization") 

def visualize_tree(tree, feature_names, save_dir='./'):
    """Create tree png using graphviz.

    Args
    ----
    tree -- scikit-learn DecsisionTree.
    feature_names -- list of feature names.
    """
    with open(save_dir+'/'+"dt.dot", 'w') as f:
        export_graphviz(tree, out_file=f,
                        feature_names=feature_names)

    command = ["dot", "-Tpng", save_dir+"/dt.dot", "-o", save_dir+"/dt.png"]
    try:
        subprocess.check_call(command)
    except:
        exit("Could not run dot, ie graphviz, to "
             "produce visualization") 

def selection_parameters_for_classfier(X,y):

    from sklearn import grid_search

    #paras={ 'n_neighbors':[1,10], 'weights':['uniform', 'distance'], 'algorithm':['auto', 'ball_tree','kd_tree', 'brute'], 'leaf_size':[20,50]}
    #knn = KNeighborsClassifier()

    #naive_bayes
    #nbg = GaussianNB()
    #nbm = MultinomialNB()
    #nbb = BernoulliNB()

    #decision tree
    #paras={ 'criterion':['gini','entropy'], 'splitter':['random', 'best'], 'max_features':[None, 'auto','sqrt', 'log2'], 'min_samples_split':[1,10]}
    #dtree = DecisionTreeClassifier()

    #random forest
    #rforest = RandomForestClassifier()
    #paras={ 'n_estimators':[2,15], 'criterion':['gini','entropy'], 'max_features': ['auto','sqrt', 'log2'], 'min_samples_split':[1,10]}

    #svm
    svmm = svm.SVC()
    paras={'kernel':['rbf','linear','poly']}


    clt =grid_search.GridSearchCV(svmm, paras, cv=5)
    clt.fit(X,y)
    print (clt)
    #print (clt.get_params())
    print (clt.set_params())
    print (clt.score(X,y))

    #scores = cross_val_score(clt,X,y,cv=10)
    #print("Accuracy: %0.2f (+/- %0.2f)" % (scores.mean(), scores.std() * 2))


#this is to get score using cross_validation 

def test_pickle_version_no_warning_is_issued_with_non_sklearn_estimator():
    iris = datasets.load_iris()
    tree = TreeNoVersion().fit(iris.data, iris.target)
    tree_pickle_noversion = pickle.dumps(tree)
    try:
        module_backup = TreeNoVersion.__module__
        TreeNoVersion.__module__ = "notsklearn"
        assert_no_warnings(pickle.loads, tree_pickle_noversion)
    finally:
        TreeNoVersion.__module__ = module_backup 

def test_pickle_version_warning_is_issued_when_no_version_info_in_pickle():
    iris = datasets.load_iris()
    # TreeNoVersion has no getstate, like pre-0.18
    tree = TreeNoVersion().fit(iris.data, iris.target)

    tree_pickle_noversion = pickle.dumps(tree)
    assert_false(b"version" in tree_pickle_noversion)
    message = pickle_error_message.format(estimator="TreeNoVersion",
                                          old_version="pre-0.18",
                                          current_version=sklearn.__version__)
    # check we got the warning about using pre-0.18 pickle
    assert_warns_message(UserWarning, message, pickle.loads,
                         tree_pickle_noversion) 

def test_pickle_version_warning_is_issued_upon_different_version():
    iris = datasets.load_iris()
    tree = TreeBadVersion().fit(iris.data, iris.target)
    tree_pickle_other = pickle.dumps(tree)
    message = pickle_error_message.format(estimator="TreeBadVersion",
                                          old_version="something",
                                          current_version=sklearn.__version__)
    assert_warns_message(UserWarning, message, pickle.loads, tree_pickle_other) 

def test_pickle_version_warning_is_not_raised_with_matching_version():
    iris = datasets.load_iris()
    tree = DecisionTreeClassifier().fit(iris.data, iris.target)
    tree_pickle = pickle.dumps(tree)
    assert_true(b"version" in tree_pickle)
    tree_restored = assert_no_warnings(pickle.loads, tree_pickle)

    # test that we can predict with the restored decision tree classifier
    score_of_original = tree.score(iris.data, iris.target)
    score_of_restored = tree_restored.score(iris.data, iris.target)
    assert_equal(score_of_original, score_of_restored) 

def baselines(n):
    td, vd, ts = data_loader.load_data(n)
    classifiers = [
        sklearn.svm.SVC(C=1000),
        sklearn.svm.SVC(kernel="linear", C=0.1),
        sklearn.neighbors.KNeighborsClassifier(1),
        sklearn.tree.DecisionTreeClassifier(),
        sklearn.ensemble.RandomForestClassifier(max_depth=10, n_estimators=500, max_features=1),
        sklearn.neural_network.MLPClassifier(alpha=1, hidden_layer_sizes=(500, 100))
    ]
    for clf in classifiers:
        clf.fit(td[0], td[1])
        print "\n{}: {}".format(type(clf).__name__, round(clf.score(vd[0], vd[1])*100, 2)) 

def transfer(n):
    td, vd, ts = data_loader.load_data(n, abstract=True, expanded=expanded)
    classifiers = [
        #sklearn.svm.SVC(),
        #sklearn.svm.SVC(kernel="linear", C=0.1),
        #sklearn.neighbors.KNeighborsClassifier(1),
        #sklearn.tree.DecisionTreeClassifier(),
        #sklearn.ensemble.RandomForestClassifier(max_depth=10, n_estimators=500, max_features=1),
        sklearn.neural_network.MLPClassifier(alpha=1.0, hidden_layer_sizes=(300,), max_iter=500)
    ]
    for clf in classifiers:
        clf.fit(td[0], td[1])
        print "\n{}: {}".format(type(clf).__name__, round(clf.score(vd[0], vd[1])*100, 2)) 

def reset(self):
        Estimator.reset(self)
        self.savedValues=None
        if self.estimatorType=='tree':
            self.tree=None
        else:
            self.policyParams=None 

def estimate(self, query, logged_ranking, new_ranking, logged_value):
        currentValue=None
        if self.savedValues is not None:
            currentValue=self.savedValues[query]
        else:
            allFeatures=self.loggingPolicy.dataset.features[query][new_ranking,:]
        
            if new_ranking.size < self.rankingSize:
                emptyPad=scipy.sparse.csr_matrix((self.rankingSize-new_ranking.size, self.numFeatures), dtype=numpy.float64)
                allFeatures=scipy.sparse.vstack((allFeatures, emptyPad), format="csr", dtype=numpy.float64)
            
            allFeatures=allFeatures.toarray()
            nRows, nCols = allFeatures.shape
            size=nRows*nCols
            currentFeatures=numpy.reshape(allFeatures, (1,size))

            if self.estimatorType=='tree':
                currentValue=self.tree.predict(currentFeatures)[0]
            else:
                currentValue=numpy.dot(currentFeatures, self.policyParams)[0]
        
            del allFeatures
            del currentFeatures
            
        self.updateRunningAverage(currentValue)
        return self.runningMean 

def __init__(self, ranking_size, logging_policy, target_policy, estimator_type):
        Estimator.__init__(self, ranking_size, logging_policy, target_policy)
        self.name = 'Direct_'+estimator_type
        self.estimatorType = estimator_type
        self.numFeatures=self.loggingPolicy.dataset.features[0].shape[1]
        self.hyperParams={'alpha': (numpy.logspace(-2,1,num=4,base=10)).tolist()}
        self.treeDepths={'max_depth': list(range(3,15,3))}
        
        if self.estimatorType=='tree':
            self.tree=None
        else:
            self.policyParams=None
            
        #This member is set on-demand by estimateAll(...)
        self.savedValues=None 

def get_fpr_tpr(clt, X, y):

    random_state = np.random.RandomState(0)
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.25 , random_state = 0)

    #from sklearn import tree
    #clt = tree.DecisionTreeClassifier( criterion='entropy', min_samples_leaf=2, min_samples_split=2, random_state=None, splitter='best')
    clt = clt.fit(X_train,y_train)
    #from sklearn.externals.six import StringIO
    #with open("iris_plus.dot", 'w') as f:
    #     f = tree.export_graphviz(clt, out_file=f)

    y_pred = clt.predict(X_test)

    #accuracy score
    _accuracy_score = accuracy_score(y_test, y_pred)

    print ("Accuracy score {}".format(_accuracy_score))

    #roc curve
    probas_ = clt.predict_proba(X_test)
    #print (probas_)
    #draw_confusion_matrix(y_test,y_pred)

    #print probas_
    fpr, tpr, thresholds = roc_curve(y_test, probas_[:, 1])
    #print (fpr, tpr,thresholds)
    roc_auc = auc(fpr, tpr)
    print ("Area under the ROC curve : %f" % roc_auc)

    return fpr, tpr , roc_auc


# this is used to draw 

def train_and_test(X,y):

    #KNN
    knn = KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski', n_neighbors=5, p=2, weights='uniform')

    #naive-bayees
    nbbern = BernoulliNB()

    #decision tree
    dtree = DecisionTreeClassifier( criterion='gini', min_samples_leaf=4, min_samples_split=2, random_state=None, splitter='best')

    #random forest
    rforest = RandomForestClassifier(bootstrap=True, criterion='gini', max_depth=None, max_features='auto',  min_samples_leaf=1, min_samples_split=2, n_estimators=10, n_jobs=1, oob_score=False, random_state=3)

    #svm
    svmrbf= svm.SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,  kernel='rbf', max_iter=-1, probability=False, random_state=None,
shrinking=True, tol=0.001, verbose=False)


    get_scroe_using_cv(knn, X, y)
    get_scroe_using_cv(nbbern, X, y)
    get_scroe_using_cv(dtree, X, y)
    get_scroe_using_cv(rforest, X, y)
    get_scroe_using_cv(svmrbf, X, y)
    print ("\n")

######################################################################

#this is to draw the Roc curve example by splitting the dataset
#just want a figure to make it more beautiful 

def my_get_fp_fn_CV(X_original,y):

    #generate classfiers
    knn = KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski', n_neighbors=5, p=2, weights='uniform')

    #decision tree
    dtree = DecisionTreeClassifier( criterion='gini', min_samples_leaf=4, min_samples_split=2, random_state=None, splitter='best')

    #naive
    #nbbern = BernoulliNB()

    #random forest
    rforest = RandomForestClassifier(bootstrap=True, criterion='gini', max_depth=None, max_features='auto',  min_samples_leaf=1, min_samples_split=2, n_estimators=10, n_jobs=1, oob_score=False, random_state=3)

    #svm
    svmrbf= svm.SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,  kernel='rbf', max_iter=-1, probability=True, random_state=None,
shrinking=True, tol=0.001, verbose=False)

    #reduce the size
    #X = SelectKBest(f_classif, k=80).fit_transform(X_original,y)
    skb = SelectKBest(f_classif, k=80).fit(X_original,y)
    X = skb.fit_transform(X_original,y)

    print ("KNN")
    my_get_fp_fn_inter(knn,X,y)
    print ("DTree")
    my_get_fp_fn_inter(dtree,X,y)
    print ("rforest")
    my_get_fp_fn_inter(rforest,X,y)
    #print ("naive bayes")
    #my_get_fp_fn_inter(nbbern,X,y)
    print ("SVMrbf")
    my_get_fp_fn_inter(svmrbf,X,y) 

def test_pickle_version_warning_is_not_raised_with_matching_version():
    iris = datasets.load_iris()
    tree = DecisionTreeClassifier().fit(iris.data, iris.target)
    tree_pickle = pickle.dumps(tree)
    assert b"version" in tree_pickle
    tree_restored = assert_no_warnings(pickle.loads, tree_pickle)

    # test that we can predict with the restored decision tree classifier
    score_of_original = tree.score(iris.data, iris.target)
    score_of_restored = tree_restored.score(iris.data, iris.target)
    assert_equal(score_of_original, score_of_restored) 

def dtree(frame, target, criterion = 'gini', depth = None, sample = 0.01, ratio = 0.15):
    tree = DecisionTreeClassifier(
        criterion = criterion,
        min_samples_leaf = sample,
        max_depth = depth,
    )

    tree.fit(frame.fillna(-1), target)

    dot_string = tree_to_dot(tree, frame.columns.values, high_light = ratio)

    dot_to_img(dot_string, file = target.name + '.png') 

def decode(cls, obj):
        import sklearn.tree

        init_args = obj['init_args']

        state = obj['state']

        # Add max_depth for backwards compatibility with PSC 1.2
        # Previous version did not set the max_depth in the state when calling __getstate__
        # https://github.com/scikit-learn/scikit-learn/blob/51a765acfa4c5d1ec05fc4b406968ad233c75162/sklearn/tree/_tree.pyx#L615

        # and has been added in sklearn 0.18 to be used in both __getstate__ and __setstate__
        # https://github.com/scikit-learn/scikit-learn/blob/ef5cb84a805efbe4bb06516670a9b8c690992bd7/sklearn/tree/_tree.pyx#L649
        
        # Older models will not have the max_depth in their stored state, such that a key error is raised.
        # the max_depth is only used in the decision path method, which we don't currently use
        # and is used to init an np array of zeros in version 0.18:
        # https://github.com/scikit-learn/scikit-learn/blob/ef5cb84a805efbe4bb06516670a9b8c690992bd7/sklearn/tree/_tree.pyx#L926
        # https://github.com/scikit-learn/scikit-learn/blob/ef5cb84a805efbe4bb06516670a9b8c690992bd7/sklearn/tree/_tree.pyx#L991
        state['max_depth'] = state.get('max_depth', 0)

        t = sklearn.tree._tree.Tree(*init_args)

        t.__setstate__(state)

        return t 

def encode(cls, obj):
        import sklearn.tree
        assert type(obj) == sklearn.tree._tree.Tree

        init_args = obj.__reduce__()[1]
        state = obj.__getstate__()

        return {
            '__mlspl_type': [type(obj).__module__, type(obj).__name__],
            'init_args': init_args,
            'state': state
        } 

def decisiontree_on_fold(feature_sets, train, test, y, y_all, X, dim, dimsum, learn_options):
    '''
    DecisionTreeRegressor from scikitlearn.
    '''
    clf = tree.DecisionTreeRegressor()
    clf.fit(X[train], y[train][:, 0])
    y_pred = clf.predict(X[test])[:, None]
    return y_pred, clf 

def test_pickle_version_no_warning_is_issued_with_non_sklearn_estimator():
    iris = datasets.load_iris()
    tree = TreeNoVersion().fit(iris.data, iris.target)
    tree_pickle_noversion = pickle.dumps(tree)
    try:
        module_backup = TreeNoVersion.__module__
        TreeNoVersion.__module__ = "notsklearn"
        assert_no_warnings(pickle.loads, tree_pickle_noversion)
    finally:
        TreeNoVersion.__module__ = module_backup 

def test_pickle_version_warning_is_issued_when_no_version_info_in_pickle():
    iris = datasets.load_iris()
    # TreeNoVersion has no getstate, like pre-0.18
    tree = TreeNoVersion().fit(iris.data, iris.target)

    tree_pickle_noversion = pickle.dumps(tree)
    assert b"version" not in tree_pickle_noversion
    message = pickle_error_message.format(estimator="TreeNoVersion",
                                          old_version="pre-0.18",
                                          current_version=sklearn.__version__)
    # check we got the warning about using pre-0.18 pickle
    assert_warns_message(UserWarning, message, pickle.loads,
                         tree_pickle_noversion) 

def test_pickle_version_warning_is_issued_upon_different_version():
    iris = datasets.load_iris()
    tree = TreeBadVersion().fit(iris.data, iris.target)
    tree_pickle_other = pickle.dumps(tree)
    message = pickle_error_message.format(estimator="TreeBadVersion",
                                          old_version="something",
                                          current_version=sklearn.__version__)
    assert_warns_message(UserWarning, message, pickle.loads, tree_pickle_other) 

def tree_to_dot(tree, features, high_light = 0.15):
    from io import StringIO
    from sklearn.tree import _tree

    out = StringIO()
    tree_ = tree.tree_

    features = np.array([
        features[i] if i != _tree.TREE_UNDEFINED else "undefined!"
        for i in tree_.feature
    ])

    out.write('digraph Tree {\n')
    out.write('edge [fontname="FangSong"];\n')
    out.write('node [shape=box];\n')

    def recurse(node, parent = None, label = None):
        sample = tree_.n_node_samples[node]
        bad_rate = tree_.value[node][0,1] / sample

        out.write('{} [label="'.format(node))

        out.write('bad rate: {:.2%}\n'.format(bad_rate))
        out.write('sample: {:.2%}\n'.format(sample / tree_.n_node_samples[0]))

        # end of label
        out.write('"')

        if bad_rate > high_light:
            out.write(', color="red"')

        # end of node
        out.write('];\n')

        if tree_.feature[node] != _tree.TREE_UNDEFINED:
            name = features[node]
            threshold = tree_.threshold[node]
            recurse(tree_.children_left[node], node, '{} <= {:.2f}'.format(name, threshold))
            recurse(tree_.children_right[node], node, '{} > {:.2f}'.format(name, threshold))

        if parent is not None:
            out.write('{} -> {} [label="{}"];\n'.format(parent, node, label))

    recurse(0, None)

    out.write('}')
    s = out.getvalue()
    out.close()
    return s 

def final_train_and_test_after_preparation(X_original,y):

    #KNN
    knn = KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski', n_neighbors=5, p=2, weights='uniform')

    #decision tree
    dtree = DecisionTreeClassifier( criterion='entropy', min_samples_leaf=4, min_samples_split=2, random_state=None, splitter='best')

    #random forest
    rforest = RandomForestClassifier(bootstrap=True, criterion='gini', max_depth=None, max_features='auto',   min_samples_leaf=1, min_samples_split=2, n_estimators=10, n_jobs=1, oob_score=False, random_state=3)

    #svm
    svmrbf= svm.SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,  kernel='rbf', max_iter=-1, probability=True, random_state=None,
shrinking=True, tol=0.001, verbose=False)

    #naive bayes
    #nbb = BernoulliNB()

    X = SelectKBest(f_classif, k=80).fit_transform(X_original,y)
    #X = X_original
    print (X.shape)
    #get_scroe_using_cv(knn, X, y)
    #get_scroe_using_cv(dtree, X, y)
    #get_scroe_using_cv(rforest, X, y)
    #get_scroe_using_cv(svmrbf, X, y)

    fpr_knn, tpr_knn, auc_knn = get_fpr_tpr(knn, X, y)
    fpr_dtree, tpr_dtree, auc_dtree = get_fpr_tpr(dtree, X, y)
    fpr_rforest, tpr_rforest, auc_rforest = get_fpr_tpr(rforest, X, y)
    fpr_svmrbf, tpr_svmrbf ,auc_svmrbf= get_fpr_tpr(svmrbf, X, y)
    #fpr_nbb, tpr_nbb ,auc_nbb= get_fpr_tpr(nbb, X, y)

    plt.clf()
    plt.plot(fpr_svmrbf, tpr_svmrbf, 'y.--', label ='SVM AUC=%0.4f'% auc_svmrbf)
    plt.plot(fpr_knn, tpr_knn, 'r^--', label='KNN AUC=%0.4f' %auc_knn)
    plt.plot(fpr_dtree, tpr_dtree, 'b>--', label ='D.Tree AUC=%0.4f'% auc_dtree)
    plt.plot(fpr_rforest, tpr_rforest, 'go--', label ='R.Forest AUC=%0.4f'% auc_rforest)
    #plt.plot(fpr_nbb, tpr_nbb, 'c*--', label ='Random Forest auc=%0.4f'% auc_nbb)


    plt.plot([0, 1], [0, 1], 'k--')
    plt.xlim([-0.02, 1.02])
    plt.ylim([-0.02, 1.02])
    plt.xlabel('FPR(False Positive Rate)',fontsize=20)
    plt.ylabel('TPR(True Positive Rate)',fontsize=20)
    #plt.title('Receiver operating characteristic ')
    plt.legend(loc="lower right")
    plt.tight_layout()
    plt.grid()
    plt.show()

    del X
    del y

##################################DRAW P-R  CURVE#######################################################
############3  this is the precisio and recall curve 

def estimateAll(self, metric=None):
        if self.savedValues is not None:
            return
            
        self.savedValues=[]
        numQueries=len(self.loggingPolicy.dataset.docsPerQuery)
        for query in range(numQueries):
            newRanking=self.targetPolicy.predict(query, self.rankingSize)
            allFeatures=self.loggingPolicy.dataset.features[query][newRanking,:]
        
            if newRanking.size < self.rankingSize:
                emptyPad=scipy.sparse.csr_matrix((self.rankingSize-newRanking.size, self.numFeatures), dtype=numpy.float64)
                allFeatures=scipy.sparse.vstack((allFeatures, emptyPad), format="csr", dtype=numpy.float64)
            
            allFeatures=allFeatures.toarray()
            nRows, nCols = allFeatures.shape
            size=nRows*nCols
            currentFeatures=numpy.reshape(allFeatures, (1,size))
        
            currentValue=None
            if self.estimatorType=='tree':
                currentValue=self.tree.predict(currentFeatures)[0]
            else:
                currentValue=numpy.dot(currentFeatures, self.policyParams)[0]
            
            low=None
            high=None
            if metric is not None:
                low=metric.getMin(newRanking.size)
                high=metric.getMax(newRanking.size)
                
            if low is not None:
                currentValue = max(currentValue, low)
            if high is not None:
                currentValue = min(currentValue, high)

            if currentValue > 1.0 or currentValue < 0.0:
                print("Direct:estimateAll [LOG] estimate %0.3f " % (currentValue), flush=True)

            del allFeatures
            del currentFeatures
            
            self.savedValues.append(currentValue)
            
            if query%100==0:
                print(".", end="", flush=True)
                
        print("")
        print("Direct:estimateAll [LOG] Precomputed estimates.", flush=True) 

def interpret(
        self,
        X_train,
        y_train,
        X_validation,
        y_validation,
        model_file_path,
        learner_name,
        target_name=None,
        class_names=None,
        metric_name=None,
        ml_task=None,
        explain_level=2,
    ):
        super(DecisionTreeRegressorAlgorithm, self).interpret(
            X_train,
            y_train,
            X_validation,
            y_validation,
            model_file_path,
            learner_name,
            target_name,
            class_names,
            metric_name,
            ml_task,
            explain_level,
        )
        if explain_level == 0:
            return
        try:

            viz = dtreeviz(
                self.model,
                X_train,
                y_train,
                target_name="target",
                feature_names=X_train.columns,
            )
            tree_file_plot = os.path.join(model_file_path, learner_name + "_tree.svg")
            viz.save(tree_file_plot)
        except Exception as e:
            logger.info(f"Problem when visuzalizin decision tree regressor. {str(e)}") 

def interpret(
        self,
        X_train,
        y_train,
        X_validation,
        y_validation,
        model_file_path,
        learner_name,
        target_name=None,
        class_names=None,
        metric_name=None,
        ml_task=None,
        explain_level=2,
    ):
        super(DecisionTreeAlgorithm, self).interpret(
            X_train,
            y_train,
            X_validation,
            y_validation,
            model_file_path,
            learner_name,
            target_name,
            class_names,
            metric_name,
            ml_task,
            explain_level,
        )
        if explain_level == 0:
            return
        try:
            if len(class_names) > 10:
                # dtreeviz does not support more than 10 classes
                return
            viz = dtreeviz(
                self.model,
                X_train,
                y_train,
                target_name="target",
                feature_names=X_train.columns,
                class_names=class_names,
            )
            tree_file_plot = os.path.join(model_file_path, learner_name + "_tree.svg")
            viz.save(tree_file_plot)
        except Exception as e:
            logger.info(f"Problem when visualizing decision tree. {str(e)}") 

def _predict_forest(model, X, joint_contribution=False):
    """
    For a given RandomForestRegressor, RandomForestClassifier,
    ExtraTreesRegressor, or ExtraTreesClassifier returns a triple of
    [prediction, bias and feature_contributions], such that prediction ≈ bias +
    feature_contributions.
    """

    if joint_contribution:
        biases = []
        contributions = []
        predictions = []
        
        for tree in model.estimators_:
            pred, bias, contribution = _predict_tree(tree, X, joint_contribution=joint_contribution)

            biases.append(bias)
            contributions.append(contribution)
            predictions.append(pred)
        
        
        total_contributions = []
        
        for i in range(len(X)):
            contr = {}
            for j, dct in enumerate(contributions):
                for k in set(dct[i]).union(set(contr.keys())):
                    contr[k] = (contr.get(k, 0)*j + dct[i].get(k,0) ) / (j+1)

            total_contributions.append(contr)    
            
        for i, item in enumerate(contribution):
            total_contributions[i]
            sm = sum([v for v in contribution[i].values()])
                

        
        return (np.mean(predictions, axis=0), np.mean(biases, axis=0),
            total_contributions)
    else:
        mean_pred = None
        mean_bias = None
        mean_contribution = None

        for i, tree in enumerate(model.estimators_):
            pred, bias, contribution = _predict_tree(tree, X)

            if i < 1: # first iteration
                mean_bias = bias
                mean_contribution = contribution
                mean_pred = pred
            else:
                mean_bias = _iterative_mean(i, mean_bias, bias)
                mean_contribution = _iterative_mean(i, mean_contribution, contribution)
                mean_pred = _iterative_mean(i, mean_pred, pred)

        return mean_pred, mean_bias, mean_contribution