Programme: MSc Financial Technology
Module: BUSI4620 Data Analytics and Machine Learning for FinTech
Coursework 2
The deadline for submission of the coursework is 3 pm on Thursday 8th May 2025. The coursework is 3,000 words of your writings with a ±10% tolerance. Appendices, codes, and the outputs of codes are not in the word count. Submit all of your writings, codes, and the outputs of codes on one single file: a Word document, or a PDF file from a Word document, or a PDF file that is printed/exported from a Jupyter Notebook. Do not submit a Jupyter Notebook.
Requirements:
You should use Python and the libraries taught in the module to solve the questions. Choose your parameter values or keep default parameters in the Python library.
You can use other Python libraries IN ADDITION TO what we covered in the module, but not some libraries to REPLACE what we have learned. Similarly, you can use some additional data that are not in the coursework requirements. These additional libraries and additional data are not compulsory.
Students cannot use the operations from other languages/software/tools such as Stata, R, Matlab, Excel, etc. as the answers.
Tasks:
The following questions require the attached dataset "EM.csv", which is the credit status and characteristics of some firms in the UK.
Do an EDA analysis of any three features that you choose. Present tables/figures and your discussions.
You must create a new feature 'CreditLevel', which is 1 (or 0) for the firms with 'Creditscore' higher (or lower) than the mean of 'Creditscore'. Drop 'Creditscore' and ‘Previouscreditscore’. The 'CreditLevel' should be the target variable 'Y’, and the remaining numerical data are 'X' in all of the models. Make data preprocessing when it is necessary as follows: keeping numerical data; filling missing values by feature means; doing StandardScaler to X.
Use three machine learning classification methods to predict 'CreditLevel', where one of them should be able to produce feature importance.
Using each method, complete the following tasks: Run a cross-validation to calculate the mean and standard deviation of the accuracy; Predict the target based on your X_test data; Use the test set to obtain a classification report; Draw a plot of the confusion matrix and a ROC plot.
Present an evaluation of these methods based on the outputs.
Using one method that can produce feature importance, draw a plot of feature importance.
Choose one of the assumptions: Machine learning can/cannot predict the credit status of the data. Provide interpretation and debate based on your results and your selected literature.
Find and link your discussions to one or several sources in terms of the following qualitative forms: texts (news/posts/papers), audio, or video recordings online.
Cite references. You can use the references provided by the module or the references found by you, or a mix of both sources of references.
Marking scheme: see the marking rubric. This is an open-question coursework. There is no ‘correct or wrong’ answer. You can choose either assumption as long as that you support it with reasonable arguments and evidence. It is acceptable if the performance of your machine learning (precision, accuracy etc.) is not high.
Guidance:
All of the above data operations are shown in the examples of Jupyter Notebooks on Moodle for our lessons. You can copy these codes to do your coursework. Be aware of different names of variables/methods/instances/dataframes/parameters etc. when you re-use the codes.
About Harvard style. references, there are a few formats. It is acceptable to follow one of them as long as you keep your reference format consistent in your coursework.
The document lists the minimum compulsory requirements. Except for the above requirements, students can carry out other relevant ideas on their own.
EXTENSIONS
The module convenor cannot grant any extensions of the deadline. Students should apply for an extension to the university following a formal procedure below. There is a panel to evaluate applications. You can directly make an application, and you do not need to ask/inform. the module convenor. An application might be rejected and hence before you receive a formal extension, you should carry on your study assuming no extension. If you didn't get an extension before the deadline, you should submit your (in)completed work before the deadline in case that your application is rejected. If your application succeeds, you can update your work later before the extended deadline. An application takes time and you should make an application as soon as possible if you need it.
Please refer to the University Quality Manual (in particular Sections, 1.5 and 1.6) at
https://www.nottingham.ac.uk/academicservices/currentstudents/extenuatingcircumstances/extenuating-circumstances-procedure.aspx
to see what constitute allowable extenuating circumstances (EC). Please note that extensions cannot be awarded for computer failure or loss, loss or damage to storage media, problems and delays collecting data and so on. Keep plenty of backups and stay organised!
Other circumstances that cannot be awarded an extension include returning home, going on holidays and any quarantine requirements in other countries. For more examples of the kind of acceptable and unacceptable circumstances, please refer to the University Quality Manual (in particular Sections, 3 and 4) at:
https://www.nottingham.ac.uk/academicservices/currentstudents/extenuatingcircumstances/guidance-on-acceptable-circumstances-and-evidence-ecprocedure.aspx
Please bear in mind that all claims under the EC procedure must be in English and supported by independent, reliable documentary evidence of inability to comply with the requirements.