"Collective Intelligence Programming", Chapter 12th: Algorithm Summary

1. Bayesian classifier:

Purpose: Classification, example: spam filtering.

is a typical supervisory algorithm. Need to be trained to classify later.

The most important aspect of Bayesian classification is the extraction of features, which transforms the data of training or classification into a list of features.

Advantages: Fast speed. The interpretation of the classifier's actual learning condition is relatively simple.

Defect: Unable to process the result of a change based on a feature combination.

2. Decision Tree Classifier

The decision tree is constructed from the root.

Depending on the characteristics of the split, the decision tree construction of the impact is very large.

"Collective Intelligence Programming", Chapter 12th: Algorithm Summary