Metocean Prediction using Hadoop, Spark & R

This project is the development of an analysis system for historical Metocean Data. It is a single page reactive web application with shiny web UI package of R containing forecasting model, ARIMA and two ML algorithms, Linear Regression and H2O AutoML developed with R for the variables of Metocean data stored in HDFS of a virtual Hadoop cluster and spark is integrated to make the computations happen in-memory. The predictions is compared to the actual data to see its correctness with RMSE. Performance difference of the application deployed on desktop and on the server is also discussed. The application performs better when running in the server than on desktop.