Grab the slides:
slides.com/cheukting_ho/legend-data-decision-trees

Every Monday 5pm UK time

Supervised Learning

Supervised learning is the machine learning task of learning a function that maps an input to an output based on example input-output pairs. It infers a function from labeled training data consisting of a set of training examples.

- wikipeida

  • data are labeled
  • lot's of human effort to aquired training data
  • let's of data cleaning before training

Unsupervised Learning

Unsupervised learning is a type of machine learning that looks for previously undetected patterns in a data set with no pre-existing labels and with a minimum of human supervision.

- wikipeida

  • data are not labeled
  • usually involved clustering
  • train to classify data based on how they cluster

Decision Tree

  • supervised learning
  • predictive model
  • epresent decisions and decision making
  • classification trees vs regression trees
  • no need to create dummy variables
  • in built feature selection
  • performs well with large datasets however non-robust
  • overfits easily

Let's have a look at Decision Trees at Sci-kit Learn

https://scikit-learn.org/stable/modules/tree.html

Decision Trees

 

Finding which feature(s) can help seperating one outcome from the others (making "decisions")

 

Sometimes a combination of decision trees can give a better picture of the "decision" and outcome statistically

 

  • Random forests - comprised of a large number of decision trees.
     
  • Tree boosting - build trees one at a time; each new tree helping to correct errors from the previously trained tree.

Random Forest

  • Generate a very large number of trees, with different sample of training data and features averaging out their solutions
     
  • Combining smaller trees to make the model more robust and prevent overfitting
     
  • It will take longer to train and harder to interpert the decision making

Gradient Boosting Trees

 

  • Faster than Random Forest because it is tuning how trees are built and combining trees on the go
     
  • But it is easily be mislead by the noises and reach a false optimization (local minimum)

Next week:

Boosting Algrithms

Every Monday 5pm UK time

The Legend of Data - Decision Trees

By Cheuk Ting Ho

The Legend of Data - Decision Trees

  • 285
Loading comments...

More from Cheuk Ting Ho