Parametric & non-parametric analysis of mean treatment effects of number of packets in transit in data network model

We analyse packet traffic in a data communication network model of the packet switching type using wavelets and principal component analysis (PCA) based methods. We focus our analysis on mean treatment effects for various factors, i.e. the pairs (routing cost metric, source load value), when the response metric is the network performance indicator number of packets in transit. We show through some examples that wavelets and PCA methods provide effective estimators of the network performance.