This project investigates the susceptibility of jobs to computerisation and particularly which features of a job determine the probability of computerisation. This is achieved by using Gaussian Process Classification. A set of labelled occupations is used to train and test the model and the effect of using different feature sets is explored. Feature selection in the form of greedy selection is used to find the feature set that achieves the best classification and thereby find the features that are most significant when determining if a job can be computerised. It is concluded that the most important feature is Originality and the best feature set for classifying the data in this project consists of Originality and Service Orientation. Furthermore, experiments are performed using linear embedding methods for feature learning. However, these experiments fail to prove that better classification can be achieved using this method.
Other Recent Journal Article / Working Papers
Vaccine nationalism and internationalism: perspectives of COVID-19 vaccine trial participants in the United Kingdom
NIMble innovation—a networked model for public antibiotic trials
Contrasting responses of large carnivores to land use management across an Asian montane landscape in Iran
Catalytic Synergy Using Al(III) and Group 1 Metals to Accelerate Epoxide and Anhydride Ring-Opening Copolymerizations
Empirically grounded technology forecasts and the energy transition
Temporal teleportation with pseudo-density operators: How dynamics emerges from temporal entanglement