Problem

• Identify Patterns in the dataset using standardized approach
• A general understanding of traffic scenarios on a given environment type
• Consideration of different traffic and environmental attributes
• Generalization by Virtual Indexing
• Unsupervised in fashion

Virtual Indexing

• A statistical approach to create ranges of different traffic and self attributes
• An iterative process of finding out an optimal split in the data, which has least variance
• Each attribute is checked for ‘goodness of variance’
• The optimal splits are indexed

Implemented using Hierarchical Density Based Clustering (Bottom up approach)

• Dendrogram has a fixed hardcoded value for epsilon (DBSCAN Hyperparameter)
• The dendrogram is traversed until ‘min sample’ number of samples is met
• Only one hyperparameter