Climatic Data Analysis by Machine Learning using R

The project imvolves model development using R software ([login to view URL] preferred over Rstudio). It is expected that a theoretical framework of the underlying models be given. Knowledge of machine learning and artificial intelligence (using R) is mandatory for this project. The deliverables of this project include the use of machine learning techniques for: 1. Visualization of the catastrophic events using heatmaps. This involves generating of a heatmap using the lattitudinal data provided against any catastrophic event. 2. Development of a hazard model which can be used to simulate catastrophic events and their frequencies for different geographical locations, while incorporating their dependencies with the rate of atmospheric buildup of carbon emissions. This will be presented in the form of heatmaps. ... The model should reproduce, e.g, previous hurricane occurences. (try use the model to reproduce catastrophes for the past, say, 100 years, and compare with the available data). 3. Building a vulnerability model to measure the severity of the damages inflicted by the catastrophic events estimated by the hazard model for different geographical locations. 4. Development of a financial model which can be used to measure the financial impact of the events simulated in (2) and the extend of damages estimated by the vulnerability model. 5. Building a parsimonious stochastic model that incorporates the three models as modules, looping climate change residuals back to the model through a sub-routine, as well as taking into consideration the dependence structures of the output from each of the models, which is expected to increase the precision of the output. Cross validation shall be an integral part of this project. Some data are provided, while others shall be obtained by the analyst.