Decision-Tree-Steroid

full implementation of decision tree with flexible impurity function and for node to be any machine learning classifier such as logistic regression, svm, as well gini and entropy impurity.

requirements

1. numpy
2. graphviz
3. scipy
4. sklearn

How to run ?

1. import important files and setup
   from Steroid.models import dt_classifier,dt_node
   from Steroid.criteria import gini,entropy
   from Steroid.impurity import gini_impurity,entropy_impurity
   from sklearn.tree import DecisionTreeClassifier
   from sklearn.linear_model import LogisticRegression
   
2. Import data in X,y as numpy array
   from sklearn.datasets import load_iris data=load_iris()
   X=data.data
   y=data.target
   
3. Initialize model
   clf=dt_classifier(no_features=[X.shape[1]],
   criteria_models={
   "Decision Tree entropy":DecisionTreeClassifier(criterion="entropy",max_depth=1,random_state=random_state),
   "Logistic Regresion":LogisticRegression(C=1,solver="lbfgs",multi_class="auto",max_iter=1000),
   }
   
4. fit data into model
   clf.fit(X_train,y_train)
   
5. predicting target
   y_pred=clf.predict(X_train)
   

View graph

graph=clf.export_graph(criteria_digits=4,show_features=False,show_value=True,show_depth=True,show_samples=True,show_leaves=True,show_criteria=True,show_model=True,show_importance=True,importance_type="percentage")
graph

some other screenshot of classification