# Portfolio

## - Bayesian Statistics: Analysis of Health Data

Here is an example of how Bayesian statistics can be used to indicate the best predictor of BMI fluctuations within a regression model.

## - Cumulative Binomial Probability with R and Shiny

The following is an illustration of how cumulative binomial probability can be calculated, and how a Shiny Web App can be used to make the analysis more intuitive.

## - Image Recognition with Keras: Convolutional Neural Networks

An example of how the VGG16 neural network can be used to classify images. In this instance, road vehicles are classified using ImageNet as a training source, and image manipulation methods with PIL are explored.

## - Kalman Filter: Modelling Time Series Shocks with KFAS in R

A common problem in modelling a time series is the presence of “shocks”, or extreme outliers in the series. Here is an illustration of how a Kalman Filter can help model for the same.

## - Modelling time series relationships between the S&P 500 and oil prices

In this example, an OLS regression model is constructed in an attempt to forecast future S&P 500 levels based on the price of Brent crude oil.

## - neuralnet: Train and Test Neural Networks Using R

Examples of how classification and regression problems can be solved using neuralnet.

## - Predicting Hotel Cancellations with ExtraTreesClassifier and Logistic Regression

Here is an example of how the ExtraTreesClassifier and Logistic Regression models can be used in Python to determine whether a potential customer will cancel their hotel booking or not.

## - Predicting Irish electricity consumption with an LSTM neural network

An LSTM model is used to forecast energy consumption of the Dublin City Council Civic Offices using data between April 2011 – February 2013.

## - SARIMA: Forecasting seasonal data with Python and R

In this example, historical weather data for Dublin, Ireland is analysed and forecasted using seasonal ARIMA methods. Specifically, auto.arima in R and Pyramid in Python is used to investigate the ideal SARIMA configuration.

## - Visualizing New York City WiFi Access with K-Means Clustering

A k-means clustering algorithm is used to analyse geographical data for free public WiFi in New York City, and the clusters are mapped geographically using nycmaps.

## - Working with panel data in R: Fixed vs. Random Effects (plm)

Here is an analysis of internet usage across both time and consumers (i.e. a panel dataset) using the plm library in R.