Significance in Machine Learning and Data Analytics Techniques on Oceanography Data

During this research experience, we were able to learn and apply several machine learning and data analysis techniques on oceanography datasets. These datasets describe 24 physical and mechanical features. Those features included temperature, intensity, longitude, and many more. The goal for this study is to utilize data mining to help oceanographers to predict and model properties as well as discover certain patterns, and physical behavior. The process of this study has started off by cleaning the datasets before applying our machine learning techniques. By that we apply interpolation and normalization to our data to be able to reconstruct it so we could have a greater overall database organization. One of our goals was to create self-organizing maps (SOMs). The purpose behind this is that it's providing data visualization techniques that help to understand high dimensional data. From the maps we can see clusters of data by which the most significant features can be extracted. The most significant features are temperature, range, and velocity. The next procedure would involve Principal Component Analysis (PCA). From our 1-D and 2-D PCA graphs we can represent a multivariate data table in order to observe clusters and/or trends. These techniques help us to predict certain patterns based on our properties. This study highlights the significance of machine learning techniques in the content of data analysis and knowledge discovery.