This project starts with an introduction to multiple linear regression using dummy variables. Using the allEffects function from the Effects package, we can visualise effects from categorical variables. The second half is an application of MLR, model building techniques and interaction effects visualisation.

This observational study aims to model and visualize the complex relationships between climatic variables in the likelikhood of rain using logistic regression and stepAIC model building functions. Some key findings are the interaction effects from temperature, air pressure, seasons and weather patterns (La Nina). The model overall performed with relatively high accuracy, specificity and sensitivity, with a high AUC (C.I going up to 91%). Unfortunately the model did not predict the weather on a given test day. But opens up areas for further analysis and application of machine learning. The data is readily available from the BoM and I am continuing to collect data to increase the reliability of the model.

An assingment in my second year course. Using different methods of classification for three vintage states of wine. I recieved a distinction for this task.

A report on GHG emissions from the world data bank. In this report, methods of multiple linear regression for emission prediction, Hierachical clustering, Kmeans and PCA for outlier detection and country clustering. PCA clustering clustered countries to 66.1% accuracy, and all unsupervised methods successfully found outliers. The MLR was not applicable as the assumption tests found non-linear relationship. A transformation of the data was needed.

This is a second assignment for a first year stats course. Analyzing probability distributions, integral transformations and basic Monte Carlo Simulations. I recieved a distinction for this piece of work.

This was an assignment tasked to me in a first year statistics course. Analyzing different probability districtions and hypothesis tests. I recieved a distinction (78%)