-
Notifications
You must be signed in to change notification settings - Fork 1
Graph Procedures Description
Abhivanth Murali edited this page Jan 9, 2024
·
11 revisions
distanceMeasure :
- "euclidean"
- "manhattan"
- "canberra"
- "cosine"
- "jaccard"
- "bray_curtis"
graphType:
- "full"
- "eps"
- "knn"
- "mknn"
parameter:
- sigma value, an Integer, used for fully connected graph calculation (specify inside string "" in neo4j, will be later converted) - integer
- k nearest neighbor value, an integer used knn, mutual knn graph (specify inside string "" in neo4j, will be later converted) - integer
- epsilon value, a double, used for epsilon graph calculation - double
- remove_column, a string of columns to be separated as coma-separated values, eg: "points,class"
label: - the label is the name of the group of nodes already present inside neo4J
- Install SimKit Plugin in Neo4j
- Procedures
- Map nodes in Neo4j from CSV
- Construct similarity matrix in Neo4j from CSV
- Construct similarity matrix in Neo4j from Neo4j Nodes
- Construct laplacian eigendecomposited matrix
- Perform K-means clustering and validate it with silhouette coefficient
- Calculate silhouette coefficient
- Calculate adjusted rand index
- Perform spectral clustering algorithm and validate it with silhouette coefficient