Classification Algorithm Based on Category Attribute’s Mathematical Expectation

The thesis introduced a classification algorithm- CAME which based on the training set’s mathematical expectation of each class attribute for unknown data. This algorithm converted the non-numerical or discrete attributes to the corresponding numerical data first, then calculate the mathematical expectation of data which belonging to different categories of numerical attributes. When a new data is needed to predict its classification, let each attribute’s mathematical expectation with existing categories as coordinate, then calculate the distance from new data attribute to various categories. The new data will belong to the category that has the shortest distance to the new data. This algorithm is not affected by attribute’s property or the number of category, and has high accuracy and good scalability.