Data Analytics for Clustering and Statistical Modeling of Oceanography Data

In this project, data mining techniques are used to test for validation in the efficiency in acquiring information about physical properties of the ocean, and bio-physical and mechanical properties of different materials, from data derived from designed experimental studies. The dataset(s) studied consist of different formulation and processing factors as inputs and different responses as outputs. The data analytics algorithms and techniques that were applied include visual assessment of clustering tendency (VAT), self-organizing maps (SOMs), and multivariate linear regression techniques (MVLR). VAT algorithm was used to help discover if there are clusters (groups) in a given dataset. SOMs was used to extract the input(s) of the most significant effect on the output responses. MVLR was applied to the dataset to estimate the associations between three different dimensions and input parameters with some of the dynamic responses in the dataset under certain conditions. This project highlights the significance and utility of data mining and statistical analysis techniques in the context of oceanography data from informatics and knowledge discovery perspective.