• Entropy controls how a DT splits data.Decision Tree reduces entropy (entropy being between 0 and 1)> Entropy formula : E = −∑(pi)log2(pi). Entropy definition: measure of impurity in a bunch of samples

• DT algorith maximizes the information Gain Information Gain formula :IG = entropy(parent)-[weighted average]entropy(children)

• Decision Boundary: axis-parallel rectangles

• Parameteres:

  • min_samples_split

• In sklearn the default criterion is slightly different from entropy ( something called gini)

See also Random Forest

• https://classroom.udacity.com/courses/ud120/
• https://www.quora.com/How-would-you-explain-Random-Forest-to-a-non-technical-person