The proximity measure of binary attributes

Binary attributes: 0 and 1. Obviously, 0 means no, 1 means

Divided into: symmetry and asymmetry

Symmetric binary attributes: two states equal importance

Asymmetry: Two states are not equally important, two take 1 (positive match) than the two take 0 (negative match) more meaningful situation

Proximity measures: Measures for dissimilarity and similarity

Q: How to characterize the differences between symmetric two-meta attributes

For:

This is a list of the two-tuple properties between Object I and Object J

Q: Indicates that the object I and Object J both take 1 of the number of attributes, and the rest are similar

P: Indicates the total number of attributes depicting the object

So dissimilarity:

Q: What about the dissimilarity of the asymmetric two-dollar attribute?

Answer: Positive match match is more meaningful, so negative match number t is negligible

So the similarity between the asymmetric two-element attribute is:, this coefficient is also called: Jaccard coefficient

When symmetric and asymmetric two-tuple properties appear in the same dataset, you can use the Blend attribute method