Overview

This analysis explores the baseline data in complimentary to the descriptive report/analysis, with Non-parametric clustering analysis for households With sufficient visualization

Clustering Analysis

The non-parametric analysis below using PAM: Partitioning Around Medoids to discover similarities between households in the PSSN-II baseline dataset. PAM instead of the go to K-means clustering is used, given the prior of the outliers and substantial variance in many variables of interests, e.g. income.

Silhouette Analysis for optimal clusters

Clustering with PAM