Decisions from Data - Publications

Partial blockout experiments for a two sided marketplace

Booking.ai (Medium)

Data Science

How can a two sided marketplace run experiments that generate insights for both sides of the marketplace simultaneously? At booking.com I developed an experiment design where we randomise customer + accommodation provider combinations into control and treatment groups, instead of only customers or only accommodation providers. This design generates unique insights, and allowed us to learn the impact a feature is having on both sides of our marketplace from a single experiment.

The identification of alpha-clustered doorway states in 44,48,52-Ti using machine learning

The European Physical Journal A

Nuclear Physics

A novel experimental analysis method has been developed, making use of the continuous wavelet transform and machine learning to rapidly identify alpha-clustering in nuclei in regions of high nuclear state density. This technique was applied to resonant scattering measurements of the 4-He(40,44,48-Ca, alpha) resonant reactions, allowing the 𝛼-cluster structure of 44,48,52-Ti to be investigated. Fragmented alpha-clustering was identified in 44-Ti and 52-Ti, while the results for 48-Ti were less conclusive, but suggest no such clustering.

Increasing the sensitivity of experiments with rank transformation

Booking.ai (Medium)

Data Science

When running online controlled experiments (A/B testing), ensuring that your experiment is sensitive enough to changes in your decision making metric is crucial. However, if your metric is very noisy, this can mean running your A/B tests for a long time, slowing down development velocity. In this project I developed a method to increase the sensitivity of A/B tests, and therefore reduce experiment runtimes, by applying a simple rank transforming on your metric.

Extracting the spectral signature of α clustering in 44,48,52-Ti using a continuous wavelet transform

Physical Review C

Nuclear Physics

A novel technique has been developed, making use of the continuous wavelet transform to identify α-clustered states from resonant scattering measurements in regions of high nuclear state density. This technique expedites the investigation of α clustering in medium mass and heavy nuclei where the role that α clustering plays in the structure of the nucleus is poorly understood. Here, we report the application of this technique to measurements of the 4-He(40, 44, 48-Ca, α) resonant reactions, leading to an assessment of the α-cluster structure of 44, 48, 52-Ti. Clustering involving α particles was identified in 44-Ti and 52-Ti but was observed to break down in 48-Ti. The implications of these results are discussed.

Alpha clustering in Ti isotopes: 40,44,48-Ca + Alpha resonant scattering

EPJ Web of Conferences

Nuclear Physics

Measurements were made of the 4-He(40,44,48-Ca,α) resonant scattering reactions at 180 degrees and up to E ∼11.5 MeV, using the Thick Target Inverse Kinematics technique. These measurements are discussed, with a focus on assessing their usefulness for investigating α-clustering in medium mass 44,48,52-Ti nuclei.

Alpha clustering in 18-F

Journal of Physics: Conference Series

Nuclear Physics

We review some of the key experimental and theoretical studies of α-clustering in 18-F. Particular attention is given to the 4p-2h nature of such α-clustered states, and the interaction between the holes and clusters is examined in terms of both weak and strong coupling regimes. The experimental work focuses on α-transfer spectroscopy and α resonant scattering as tools for investigating α-clustering.

Energy Levels of 18-F from the 14-N + Alpha Resonant Reaction

Physical Review C

Nuclear Physics

Measurements were made of the differential cross section of the resonant reactions 4-He(14-N,α) and, 4-He(14-N,d)16-O with the intention of investigating the compound nucleus 18-F. These measurements were performed in inverse kinematics at a center-of-mass scattering angle of θ =180 degrees by using a thick 4-He gas target and a 14-N beam. Data were recorded which covered 18-F excitation energies from 5.5–16 MeV. An R-matrix analysis was performed on the data up to 9 MeV, and the energies, spins, parities, and partial widths were extracted. Nine new states have been identified in 18-F based on this analysis between 8.326 and 8.915 MeV.