Data Reduction Network

Multidimensional categorical data is widespread but not easily visualized using standard methods. For example, questionnaire (e.g. survey) generally consists of questions with responses (e.g., yes/no, hate/dislike/neutral/like/love). Thus, a 10 questions, each five mutually exclusive responses, gives dataset $5^{10}$ possible observations, an amount that would be hard to reasonably collect. Hence, this type necessarily sparse. Popular methods handling include one-hot encoding (which exacerbates the dimensionality problem) and enumeration, which applies unwarranted potentially misleading notional order data. To address this, we introduce novel visualization method named Data Reduction Network (DRN). Using network-graph structure, DRN denotes feature as node interrelationships between nodes denoted by weighted edges. The graph statistically reduced reveal strongest or weakest path-wise relationships features reduce visual clutter. A key advantage it does “lose” features, rather represents across entire set without eliminating weaker features. Indeed, representation can inverted so instead visualizing interrelationships, surfaced. powerful tool for multi-dimensional in particular derived from surveys questionaires.