Complex Traffic Network Analysis Method Based on a Multiscale Aggregation Model
Algorithm 1
Improved spectral clustering algorithm flow.
Input: sample points and the number of clusters
1. Calculate the similarity matrix of ;
2. Based on the similarity matrix , calculate the weighting matrix considering the weight influence factors of location, distance, road grade, and traffic congestion degree;
3. Calculate the degree matrix ;
4. Calculate the Laplace matrix ;
5. Calculate the eigenvalues of Laplace matrix P and sort the eigenvalues from small to large;
6. According to the two-dimensional decision diagram in 4.2, select the central nodes and get the number of categories ;
7. Take the first m eigenvalues of Laplace matrix and calculate the eigenvectors of the first m eigenvalues ;
8. Form the above m column vectors into a matrix ;
9. Let be the vector of line of, where ;
10. For , is sequentially united so that ;
11. Clustering new sample point into cluster using -means algorithm;
