## 2018

Accurately predicting when and where ambulance call-outs occur can reduce response times and ensure the patient receives urgent care sooner. Here we present a novel method for ambulance demand prediction using Gaussian Process Regression (GPR) in time and geographic space. The method exhibits superior accuracy to MEDIC, a method which has been used in industry. The use of GPR has additional benefits such as the quantification of uncertainty with each prediction, the choice of kernel functions to encode prior knowledge and the ability to capture spatial correlation. Measures to increase the utility of GPR in the current context, with large training sets and a Poisson-distributed output, are outlined.

## 2016

The purpose of this research is to determine if patient survival and mechanisms of right ventricular failure in pulmonary hypertension could be predicted by using supervised machine learning of three-dimensional patterns of systolic cardiac motion. A machine learning survival model that uses three-dimensional cardiac motion predicts outcome independent of conventional risk factors in patients with newly diagnosed pulmonary hypertension.

## 2015

Journal of Wine Economics

## Machine Learning in Fine Wine Price Prediction

Advanced machine learning techniques like Gaussian process regression and multi-task learning are novel in the area of wine price prediction; previous research in this area being restricted to parametric linear regression models when predicting wine prices. The paper, using historical price data of the 100 wines in the Liv-Ex 100 index, makes 3 main contributions to the field. Firstly, a clustering of the wines into two distinct collections based on autocorrelation. Secondly, an implementation of Gaussian process regression on these wines with predictive accuracy surpassing both the trivial and simple ARMA and GARCH time series prediction benchmarks. Lastly, an implementation of an algorithm which performs multi-task feature learning with kernels on the wine returns as an extension to our optimal Gaussian process regression model. Using the optimal covariance kernel from Gaussian process regression, we achieve predictive results which are comparable to that of Gaussian process regression. Altogether, our research suggests that there is potential in using advanced machine learning techniques in wine price prediction.

## 2013

Computational Economics

## Multiple Kernel Learning with Fisher kernels for High Frequency Currency Prediction

Financially motivated kernels based on EURUSD currency data are constructed from limit order book volumes, commonly used technical analysis methods and canonical market microstructure models—the latter in the form of Fisher Kernels. These kernels are used through their incorporation into support vector machines (SVM) to predict the direction of price movement for the currency over multiple time horizons. Multiple Kernel Learning is used to replicate the signal combination process that trading rules embody when they aggregate multiple sources of financial information. Significant outperformance relative to both the individual SVM and benchmarks is found, along with an indication of which features are the most informative for financial prediction tasks. An average accuracy of 55% is achieved when classifying the direction of price movement into one of three categories for a 200 s predictive time horizon.

## 2010

NIPS 2010 Workshop: New Directions in Multiple Kernel Learning

## Currency Forecasting using Multiple Kernel Learning with Financially Motivated Features

Financially motivated kernels are used through their incorporation into support vector machines (SVM) to predict the direction of price movement for the currency over multiple time horizons. These kernels are based on EURUSD currency data are constructed from limit order book volumes, commonly used technical analysis methods and canonical market microstructure models—the latter in the form of Fisher Kernels. Multiple Kernel Learning is used to replicate the signal combination process that trading rules embody when they aggregate multiple sources of financial information. Significant outperformance relative to both the individual SVM and benchmarks is found, along with an indication of which features are the most informative for financial prediction tasks. An average accuracy of 55% is achieved when classifying the direction of price movement into one of three categories for a 200 s predictive time horizon.

## 2010

Journal of Machine Learning Research Proceedings

## Multiple Kernel Learning on the Limit Order Book

Simple features constructed from order book data for the EURUSD currency pair are used to construct a set of kernels. These kernels are used both individually and simultaneously through the Multiple Kernel Learning (MKL) methods of SimpleMKL and the more novel LPBoostMKL to train multiclass Support Vector Machines to predict the direction of future price movements. The kernel methods outperform a trend following benchmark both in their predictive ability and when used in a simple trading rule. Furthermore, the kernel weightings selected by the MKL techniques highlight which features of the EURUSD order book are the most informative for predictive tasks.

## 2009

Proceedings of the International Conference of Financial Engineering

## Machine Learning in FX Carry Basket Prediction

Neural Networks (ANN), Support Vector Machines (SVM) and Relevance Vector Machines (RVM) are used to predict daily returns for an FX carry basket. Market observable exogenous variables known to have a relationship with the basket along with lags of the basket's return are used as inputs into these methods. Combinations of these networks are used in a committee and simple trading rules based on this amalgamated output are used to predict when carry basket returns would be negative for a day and hence a trader should go short this long-biased asset. The effect of using the networks for regression to predict actual returns is compared to their use as classifiers to predict whether the following day's return would be up or down. Assuming highly conservative estimates of trading costs, over the 10.5 year (2751 trading day) rolling out of sample period investigated, improvements of 120% in MAR ratio, 110% in Sortino and 80% in Sharpe relative to the `Always In' benchmark are found. Furthermore, the extent of the maximum draw-down is reduced by 19% and the longest draw-down period is 53% shorter.

## 2012

The usage of machine learning techniques for the prediction of financial time series is investigated. Both discriminative and generative methods are considered and compared to more standard financial prediction techniques. Generative methods such as Switching Autoregressive Hidden Markov and changepoint models are found to be unsuccessful at predicting daily and minutely prices from a wide range of asset classes. Committees of discriminative techniques (Support Vector Machines (SVM), Relevance Vector Machines and Neural Networks) are found to perform well when incorporating sophisticated exogenous financial information in order to predict daily FX carry basket returns. The higher dimensionality that Electronic Communication Networks make available through order book data is transformed into simple features. These volume-based features, along with other price-based ones motivated by common trading rules, are used by Multiple Kernel Learning (MKL) to classify the direction of price movement for a currency over a range of time horizons. Outperformance relative to both individual SVM and benchmarks is found, along with an indication of which features are the most informative for financial prediction tasks. Fisher kernels based on three popular market microstructural models are added to the MKL set. Two subsets of this full set, constructed from the most frequently selected and highest performing individual kernels are also investigated. Furthermore, kernel learning is employed - optimising hyperparameter and Fisher feature parameters with the aim of improving predictive performance. Significant improvements in out-of-sample predictive accuracy relative to both individual SVM and standard MKL is found using these various novel enhancements to the MKL algorithm.