Team Members:
Context:
Somerset is a county in the Southwest of England, boasting five areas of outstanding natural beauty, three residing solely in Somerset. The Visitor Economy is worth to Somerset £1.3 Billion and 24,000 FTE jobs annually. Visit Somerset is the official destination management organisation for the county of Somerset, officially representing the Visitor Economy. Working alongside its larger providers such as Visit England and Visit Britain. Visit Somerset has by the nature of its operations possibly the largest digital platform in the county, looking after over the last 2 years over a million followers and users, through its digital platforms. This creates a huge opportunity for the long-term development of data focussed strategy, including the inception of Chat Bots, AI and Machine Learning capability, in support of programmatic marketing, personalisation, productivity enhancement, and more effective customer service values. As a prequel to these aspirations, we need to perform some data modelling based around a basic issue Visit Somerset experience around returning visitors and the lack of, illustrated as follow:
According to data taken from Google analytics, it is realised that there are relatively high new visitors compared to those returning. Visit Somerset want to use a data driven approach to aim to understand and identify the cause of less returning visitors and, ideally, to create an algorithm that could incentivise the people's behaviour towards returning
Data:
There is data available for visitor numbers from Google Analytics (7 years) and ONS.
Further is available for Visitor Behaviour
including demographics
, use of car parks, Visitor impact (NHS, Public services, pollution, forest fires, Amenities, Transport connectivity, Climate Crisis), and geographical movement
Categorical data that we might also need to investigate includes: Strikes
, Weather / Climate
, Booking time and how many (number)
, Cancellations
, Correlation to other National events
, Impact of COVID
, Impact of economic instability
The project:
Will be to construct a model which will aim to link returning visitor numbers to possible causes in a way that could be predictive for future numbers.
The project will comprise the following
- Acquisition and preliminary representation of the various forms of data required
- Preliminary statistical analysis of the data to establish meaningful correlations between possible causes and effects linking the categorical and visitor data.
- Construction, training and implementation, of a machine learning model (probably some form of RNN) to predict the returning visitor numbers.
Supervision:
The student will be supervised by Prof Budd (Bath) and Mr John Turner (Visit Somerset) with assistance on modelling from Robert Clarke (Bath) and data from Warren George (GoGetOrganised). They will have regular meetings with the supervision team.