UNIT CODE:ACFIM0002

UNIT NAME: AI, Blockchain Technology and Applications

December 2024

Overview

•     Your summative coursework represents 40% of the final mark for the unit.

•     The coursework is in the form of an essay.

•     Penalties will apply if the coursework is submitted late.

•     The coursework is a collaborative work - you should work on this as a group. You will be required to make a plagiarism statement and your submission will be tested for originality.

Coursework requirement

•     You are expected to prepare one report (maximum 3000 words) in accordance with the below requirements.

•     Select any two questions from the following three, and please note that the group project has a maximum score of 100.

Primary Data Source

TheMSCI ESG KLD STATS dataset is a comprehensive annual dataset that tracks various indicators of environmental, social, and governance (ESG) performance for publicly listed companies. This dataset, which dates back to 1991, covers abroad range of industries and includes both positive and negative factors related to corporate ESG practices.

Key variables in the dataset include detailed metrics on environmental performance, corporate governance, employee relations, human rights, product responsibility, and community impact. The ESG indicators are designed to capture how well companies manage risks and opportunities in these areas. The dataset offers a rich foundation for analysing trends in corporate responsibility and sustainability overtime.

Question 1 (50 marks, maximum 1500 words): Supervised Regression Analysis

Objective: Explore the connections between companies' environmental and governance practices. You may need to incorporate additional data sources for a more comprehensive analysis.

Methodological Framework: Each variable included in your model should be logically justified, grounded in sound economic or business principles. Additionally, ensure that your regression model includes no more than 20 independent variables.

Evaluation Metrics: Use the coefficient of determination (R²) and the Sum of Squared Errors (SSE) as your key statistical tools for evaluating the best-fitting model. These metrics will help you assess the model's explanatory power and accuracy.

Predictive Accuracy: Pay attention to the predictive capability of your models. This requires setting aside a portion of your data—ideally chosen at random—as a validation set to evaluate the accuracy of your model’spredictions.

Question 2 (50 marks, maximum 1500 words): Supervised Classification Analysis

Objective: Analyse the key  factors that  affect  companies'  employee  relations.  This requires  selecting  a  specific employee relations metric (positive or negative) and investigating how corporate characteristics, possibly supplemented by data from external sources, either enhance or undermine employee relations.

Analytical Approach: Your analysis should be grounded in supervised classification techniques, with a particular emphasis on logistic regression models or linear Support Vector Machines (SVMs). These methods are well-suited for handling binary outcome variables and provide clear decision boundaries for classification.

Question 3 (50 marks, maximum 1500 words): Textual Analysis

Objective: Create a new variable derived from textual data, such as company annual reports, analyst reports, conference calls, or government websites. Analyse how this newly constructed variable impacts firms’ environmental performance.

Methodological Pathway: This task will involve applying textual analysis or natural language processing (NLP) techniques to parse, interpret, and extract meaningful insights from textual content.

Reporting Guidelines

•     In your report, it's important to mix your technical findings with clear explanations from economics or finance, showing why certain trends or unusual findings are important.

•     Also, you need to keep a steady flow and way of citing sources in your report. It's important to stick to a well- known referencing style. like Harvard to make sure your report is trustworthy and readers can check where your information comes from.

•     You have the option to use various datasets to tackle questions similar to those in the MSCI dataset. Make sure to apply the methods specified in the Questions for consistency and relevance.

•     You need to submit your final essay, Python code and datasets.