Introduction

This project is an opportunity for you to apply your learned knowledge and skills towards real marketing problems. You will work in teams of 4–6 students to analyse data and provide insights and actionable recommendations to a company of your own choosing. You are responsible for forming your own teams. Once your team is formed, register your team by emailing Zejing Shao ([email protected]) each team member’s name, student number, and UCL email address. Remember to cc your team members in the email to keep everyone informed. The deadline for team registration is 15 November 2023. You should take the decision of team formation seriously: once it is registered after the deadline, it cannot be changed and you will be assessed as a team, meaning that everyone will receive the same mark barring extremely exceptional circumstances.

Company and Research Question

You are responsible for deciding the company/industry about which you will do your project. One piece of advice here is to be flexible: once you’ve identified the company/industry and formulated your research question, you need to find the company’s data to answer your research question, which significantly limits your choice set.

After you’ve decided the company, you need to formulate a research question and possibly some sub-questions that together inform the company of its marketing strategy. Needlessly to say, the research question should be SMART.

Your research question and its subsequent analysis should be tied to at least one of components in the 3Cs, the 4Ps, or the STP. Below are some examples of the research questions in the past:

• Is there any temporal trend in the listing prices at Airbnb? How do the trends vary across different cities? What are the actionable recommendations we can come up with based on the identified trends? These questions address the pricing component of the 4Ps and help

Airbnb make better pricing recommendation to its hosts.

Data and Research Method

You need to find and use real datasets for your research question. The data can be either secondary or primary. If you are using any datasets that are not publicly available, you need to obtain the proper authorization to access and share the data with your markers. For example, while scraping data from a website is popular nowadays, many website owners are concerned about the practice and have prohibited the use of scraping tools on their websites. In such occasions, it is important to check the website’s terms of use. Typically, a website that allows scraping will provide a public API for you to access its data. If you are using any data that are not publicly available, you must also obtain the authorization to share your data with your module leader and teaching assistants as we may need to check whether we could replicate your analysis.

For your analysis of the data, it should include both descriptive data visualization and the imple- mentation of supervised learning (eg regression) or unsupervised learning (eg clustering). Large language models such as OpenAI’s GPT models, Google’s PaLM, and Meta’s LLaMA, have received a lot of attention from both the academia and the industry recently. Studies so far suggest that they have large potential in tasks related to marketing such as sentiment analysis (Rathje et al. 2023; Wang et al. 2023; Zhang et al. 2023) and consumer simulation in conjoint analysis (Brand, Israeli, and Ngwe 2023). For this coursework, you are allowed to use such models to assist your research such as labelling sentiments of texts, but you are not allowed to use them as the main research method. An example of using AI in an assistive manner is to extract the sentiment of product reviews and then use the sentiment as a predictor in a regression model.

For the aforementioned example about Airbnb, one example of the possible research methods is visualizing the spatial distribution of prices and then regressing the prices on a set of predictors in an attempt to recover the underlying factors driving the price differences across cities overtime.

For the implementation of your analysis, you are free to use whatever computational tools you like, such as Julia, Python, and R.

Deliverables

You are required to submit a report presenting your work to the designated dropbox on Moodle in a single PDF file by Friday 15 December 2023. You should also email your module leader the codes and datasets you have used for your analysis. If the dataset is publicly available, you should provide the link to the dataset.

The writing style of your report should be business or academic formal. You should prepare your report with a target audience of business professionals that understand marketing, statistics, and the core ideas of machine learning. Your report should not include any code in the main body of the report, as this coursework is about marketing, not programming. Treating the coursework as one for your data analytics module is likely to result in a low mark. The length limit of the report is 2,500 words for the main body of the report. You’ll also need to make a slide deck of your report. Details about the slide deck are included in the slides uploaded to Moodle.

Submission instructions for the slide deck will be announced later after coordination with the

Programme Administration.

An easy way to format and structure your report is as follows:

1.  Company/industry background and an introduction to your research question

2.  Data description and summary statistics that help your readers understand how the data are useful for answering your research question.

3. Description of your research method and then its implementation.

4. Presentation of your analysis and the relevant managerial/business implications.

Assessment

The group coursework accounts for 30% of your overall assessment. Given the open-ended nature of the coursework,it is impossible to enumerate all the criteria for assessment.  However, the following list should give you a rough idea of what we are looking for:

• Whether your research question is SMART.

• Whether you’ve given a clear description and motivation for your research question and the company/industry of interest.

• Whether you’ve given a clear motivation for the dataset you’ve chosen.

• Whether you’ve given just enough information about the data that helps your reader under- stand the background of the research question.

• Whether your chosen research method is appropriate for your research question and data.

• Whether you’ve implemented the research method correctly.

• Whether you’ve presented your analysis in a clear, concise, and informative manner.

• Whether you’ve provided actionable and quantifiable recommendations that are attributable to your analysis.

Lastly, it is important to set the right expectation for your coursework: a report that does well in almost every listed point above is likely to receive a mark of 60+. A report with a mark of 70+ needs to be outstanding in at least several of the points above, which is objectively difficult to achieve given the time constraint you have for this coursework.

Some Publicly Available Datasets

Some publicly available datasets also come with publicly available codes, such as those from Kaggle.

Under no circumstances are you allowed to use any existing code posted online for your project. Doing so is considered an academic misconduct and will be escalated to the University’sMisconduct Procedure.

Hospitality and Travel

• Inside Airbnb