A key source of insights in time series data comes from multivariate trends, also called patterns, which are often reviewed forensically to understand past system behavior. Such patterns are often hard to describe and cannot be easily captured by traditional analytics approaches.
Read Original Article