Multi-viewpoints based Visual methods for Efficient Exploratory Data Analysis of Current Events and Trends

Cluster is a basic tool and technique of Exploratory Data analysis (EDA) which can be used to partition data into groups or clusters based on similarity measures. Present visualization methods and traditional visual assessment tendency (VAT), use Euclidean distance in most of the cases and cosine metrics in some more cases. Cosine metrics work more effectively on text data than Euclidean since Euclidean distance is used for numerical attributes. Cosine metrics can work better on Twitter data, although it concerns on single viewpoint which might convey less information. Hence, multi-viewpoints can be used to overcome this drawback and works effectively on Twitter datasets with multiple topics. In this study,multi-viewpoint is used as similarity metrics with traditional topic models to form clusters. Got results were compared with models like VAT, cVAT using validity indices, computational complexities, and speedup factors.