Mastering Azure Machine Learning by Christoph Körner

Author:Christoph Körner [Christoph Körner]
Language: eng
Format: epub
Tags: COM032000 - COMPUTERS / Information Technology, COM018000 - COMPUTERS / Data Processing, COM021030 - COMPUTERS / Databases / Data Mining
Publisher: Packt
Published: 2020-04-29T15:05:40+00:00

You can find tree-based bagging ensembles such as random forest, and sometimes also extra-trees, in many popular ML libraries, such as scikit-learn, Spark MLlib, ML.NET, and many others.

Optimizing classifiers with boosting rounds

Often in computer science problems, we can replace a random greedy approach with a more complex but more optimal approach. The same holds true for tree ensembles and builds the foundation for boosted tree ensembles.

The basic idea behind boosting is quite simple:

We start to train an individual model on the whole training data.

Then we compute the predictions of the model on the training data and start weighting training samples that yield a wrong result higher.

Next, we train another decision tree using the weighted training set. We then combine both decision trees into an ensemble and again predict the output classes for the weighted training set. As you might have guessed, we further increase the weights on the wrongly classified training samples for the next boosting round.

We continue this algorithm until a stopping criterion is reached.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.